Debian Codegraph @

rfc9135.html

raw
1 <!DOCTYPE html>
2 <html lang="en" class="RFC">
3 <head>
4 <meta charset="utf-8">
5 <meta content="Common,Latin" name="scripts">
6 <meta content="initial-scale=1.0" name="viewport">
7 <title>RFC 9135: Integrated Routing and Bridging in Ethernet VPN (EVPN)</title>
8 <meta content="Ali Sajassi" name="author">
9 <meta content="Samer Salam" name="author">
10 <meta content="Samir Thoria" name="author">
11 <meta content="John E Drake" name="author">
12 <meta content="Jorge Rabadan" name="author">
13 <meta content="
14
15 Ethernet VPN (EVPN) provides an extensible and flexible multihoming
16 VPN solution over an MPLS/IP network for intra-subnet connectivity
17 among Tenant Systems and end devices that can be physical or virtual.
18 However, there are scenarios for which there is a need for a dynamic
19 and efficient inter-subnet connectivity among these Tenant Systems
20 and end devices while maintaining the multihoming capabilities of
21 EVPN. This document describes an Integrated Routing and Bridging
22 (IRB) solution based on EVPN to address such requirements.
23 " name="description">
24 <meta content="xml2rfc 3.10.0" name="generator">
25 <meta content="IRB" name="keyword">
26 <meta content="inter-subnet-forwarding" name="keyword">
27 <meta content="symmetric" name="keyword">
28 <meta content="asymmetric" name="keyword">
29 <meta content="mobility" name="keyword">
30 <meta content="9135" name="rfc.number">
31 <!-- Generator version information:
32 xml2rfc 3.10.0
33 Python 3.6.13
34 appdirs 1.4.4
35 ConfigArgParse 1.4.1
36 google-i18n-address 2.4.0
37 html5lib 1.0.1
38 intervaltree 3.0.2
39 Jinja2 2.11.3
40 kitchen 1.2.6
41 lxml 4.4.2
42 pycairo 1.15.1
43 pycountry 19.8.18
44 pyflakes 2.1.1
45 PyYAML 5.4.1
46 requests 2.24.0
47 setuptools 40.5.0
48 six 1.14.0
49 WeasyPrint 52.5
50 -->
51 <link href="rfc9135.xml" rel="alternate" type="application/rfc+xml">
52 <link href="#copyright" rel="license">
53 <style type="text/css">/*
54
55 NOTE: Changes at the bottom of this file overrides some earlier settings.
56
57 Once the style has stabilized and has been adopted as an official RFC style,
58 this can be consolidated so that style settings occur only in one place, but
59 for now the contents of this file consists first of the initial CSS work as
60 provided to the RFC Formatter (xml2rfc) work, followed by itemized and
61 commented changes found necssary during the development of the v3
62 formatters.
63
64 */
65
66 /* fonts */
67 @import url('https://fonts.googleapis.com/css?family=Noto+Sans'); /* Sans-serif */
68 @import url('https://fonts.googleapis.com/css?family=Noto+Serif'); /* Serif (print) */
69 @import url('https://fonts.googleapis.com/css?family=Roboto+Mono'); /* Monospace */
70
71 @viewport {
72 zoom: 1.0;
73 width: extend-to-zoom;
74 }
75 @-ms-viewport {
76 width: extend-to-zoom;
77 zoom: 1.0;
78 }
79 /* general and mobile first */
80 html {
81 }
82 body {
83 max-width: 90%;
84 margin: 1.5em auto;
85 color: #222;
86 background-color: #fff;
87 font-size: 14px;
88 font-family: 'Noto Sans', Arial, Helvetica, sans-serif;
89 line-height: 1.6;
90 scroll-behavior: smooth;
91 }
92 .ears {
93 display: none;
94 }
95
96 /* headings */
97 #title, h1, h2, h3, h4, h5, h6 {
98 margin: 1em 0 0.5em;
99 font-weight: bold;
100 line-height: 1.3;
101 }
102 #title {
103 clear: both;
104 border-bottom: 1px solid #ddd;
105 margin: 0 0 0.5em 0;
106 padding: 1em 0 0.5em;
107 }
108 .author {
109 padding-bottom: 4px;
110 }
111 h1 {
112 font-size: 26px;
113 margin: 1em 0;
114 }
115 h2 {
116 font-size: 22px;
117 margin-top: -20px; /* provide offset for in-page anchors */
118 padding-top: 33px;
119 }
120 h3 {
121 font-size: 18px;
122 margin-top: -36px; /* provide offset for in-page anchors */
123 padding-top: 42px;
124 }
125 h4 {
126 font-size: 16px;
127 margin-top: -36px; /* provide offset for in-page anchors */
128 padding-top: 42px;
129 }
130 h5, h6 {
131 font-size: 14px;
132 }
133 #n-copyright-notice {
134 border-bottom: 1px solid #ddd;
135 padding-bottom: 1em;
136 margin-bottom: 1em;
137 }
138 /* general structure */
139 p {
140 padding: 0;
141 margin: 0 0 1em 0;
142 text-align: left;
143 }
144 div, span {
145 position: relative;
146 }
147 div {
148 margin: 0;
149 }
150 .alignRight.art-text {
151 background-color: #f9f9f9;
152 border: 1px solid #eee;
153 border-radius: 3px;
154 padding: 1em 1em 0;
155 margin-bottom: 1.5em;
156 }
157 .alignRight.art-text pre {
158 padding: 0;
159 }
160 .alignRight {
161 margin: 1em 0;
162 }
163 .alignRight > *:first-child {
164 border: none;
165 margin: 0;
166 float: right;
167 clear: both;
168 }
169 .alignRight > *:nth-child(2) {
170 clear: both;
171 display: block;
172 border: none;
173 }
174 svg {
175 display: block;
176 }
177 .alignCenter.art-text {
178 background-color: #f9f9f9;
179 border: 1px solid #eee;
180 border-radius: 3px;
181 padding: 1em 1em 0;
182 margin-bottom: 1.5em;
183 }
184 .alignCenter.art-text pre {
185 padding: 0;
186 }
187 .alignCenter {
188 margin: 1em 0;
189 }
190 .alignCenter > *:first-child {
191 border: none;
192 /* this isn't optimal, but it's an existence proof. PrinceXML doesn't
193 support flexbox yet.
194 */
195 display: table;
196 margin: 0 auto;
197 }
198
199 /* lists */
200 ol, ul {
201 padding: 0;
202 margin: 0 0 1em 2em;
203 }
204 ol ol, ul ul, ol ul, ul ol {
205 margin-left: 1em;
206 }
207 li {
208 margin: 0 0 0.25em 0;
209 }
210 .ulCompact li {
211 margin: 0;
212 }
213 ul.empty, .ulEmpty {
214 list-style-type: none;
215 }
216 ul.empty li, .ulEmpty li {
217 margin-top: 0.5em;
218 }
219 ul.ulBare, li.ulBare {
220 margin-left: 0em !important;
221 }
222 ul.compact, .ulCompact,
223 ol.compact, .olCompact {
224 line-height: 100%;
225 margin: 0 0 0 2em;
226 }
227
228 /* definition lists */
229 dl {
230 }
231 dl > dt {
232 float: left;
233 margin-right: 1em;
234 }
235 /*
236 dl.nohang > dt {
237 float: none;
238 }
239 */
240 dl > dd {
241 margin-bottom: .8em;
242 min-height: 1.3em;
243 }
244 dl.compact > dd, .dlCompact > dd {
245 margin-bottom: 0em;
246 }
247 dl > dd > dl {
248 margin-top: 0.5em;
249 margin-bottom: 0em;
250 }
251
252 /* links */
253 a {
254 text-decoration: none;
255 }
256 a[href] {
257 color: #22e; /* Arlen: WCAG 2019 */
258 }
259 a[href]:hover {
260 background-color: #f2f2f2;
261 }
262 figcaption a[href],
263 a[href].selfRef {
264 color: #222;
265 }
266 /* XXX probably not this:
267 a.selfRef:hover {
268 background-color: transparent;
269 cursor: default;
270 } */
271
272 /* Figures */
273 tt, code, pre, code {
274 background-color: #f9f9f9;
275 font-family: 'Roboto Mono', monospace;
276 }
277 pre {
278 border: 1px solid #eee;
279 margin: 0;
280 padding: 1em;
281 }
282 img {
283 max-width: 100%;
284 }
285 figure {
286 margin: 0;
287 }
288 figure blockquote {
289 margin: 0.8em 0.4em 0.4em;
290 }
291 figcaption {
292 font-style: italic;
293 margin: 0 0 1em 0;
294 }
295 @media screen {
296 pre {
297 overflow-x: auto;
298 max-width: 100%;
299 max-width: calc(100% - 22px);
300 }
301 }
302
303 /* aside, blockquote */
304 aside, blockquote {
305 margin-left: 0;
306 padding: 1.2em 2em;
307 }
308 blockquote {
309 background-color: #f9f9f9;
310 color: #111; /* Arlen: WCAG 2019 */
311 border: 1px solid #ddd;
312 border-radius: 3px;
313 margin: 1em 0;
314 }
315 cite {
316 display: block;
317 text-align: right;
318 font-style: italic;
319 }
320
321 /* tables */
322 table {
323 width: 100%;
324 margin: 0 0 1em;
325 border-collapse: collapse;
326 border: 1px solid #eee;
327 }
328 th, td {
329 text-align: left;
330 vertical-align: top;
331 padding: 0.5em 0.75em;
332 }
333 th {
334 text-align: left;
335 background-color: #e9e9e9;
336 }
337 tr:nth-child(2n+1) > td {
338 background-color: #f5f5f5;
339 }
340 table caption {
341 font-style: italic;
342 margin: 0;
343 padding: 0;
344 text-align: left;
345 }
346 table p {
347 /* XXX to avoid bottom margin on table row signifiers. If paragraphs should
348 be allowed within tables more generally, it would be far better to select on a class. */
349 margin: 0;
350 }
351
352 /* pilcrow */
353 a.pilcrow {
354 color: #666; /* Arlen: AHDJ 2019 */
355 text-decoration: none;
356 visibility: hidden;
357 user-select: none;
358 -ms-user-select: none;
359 -o-user-select:none;
360 -moz-user-select: none;
361 -khtml-user-select: none;
362 -webkit-user-select: none;
363 -webkit-touch-callout: none;
364 }
365 @media screen {
366 aside:hover > a.pilcrow,
367 p:hover > a.pilcrow,
368 blockquote:hover > a.pilcrow,
369 div:hover > a.pilcrow,
370 li:hover > a.pilcrow,
371 pre:hover > a.pilcrow {
372 visibility: visible;
373 }
374 a.pilcrow:hover {
375 background-color: transparent;
376 }
377 }
378
379 /* misc */
380 hr {
381 border: 0;
382 border-top: 1px solid #eee;
383 }
384 .bcp14 {
385 font-variant: small-caps;
386 }
387
388 .role {
389 font-variant: all-small-caps;
390 }
391
392 /* info block */
393 #identifiers {
394 margin: 0;
395 font-size: 0.9em;
396 }
397 #identifiers dt {
398 width: 3em;
399 clear: left;
400 }
401 #identifiers dd {
402 float: left;
403 margin-bottom: 0;
404 }
405 /* Fix PDF info block run off issue */
406 @media print {
407 #identifiers dd {
408 float: none;
409 }
410 }
411 #identifiers .authors .author {
412 display: inline-block;
413 margin-right: 1.5em;
414 }
415 #identifiers .authors .org {
416 font-style: italic;
417 }
418
419 /* The prepared/rendered info at the very bottom of the page */
420 .docInfo {
421 color: #666; /* Arlen: WCAG 2019 */
422 font-size: 0.9em;
423 font-style: italic;
424 margin-top: 2em;
425 }
426 .docInfo .prepared {
427 float: left;
428 }
429 .docInfo .prepared {
430 float: right;
431 }
432
433 /* table of contents */
434 #toc {
435 padding: 0.75em 0 2em 0;
436 margin-bottom: 1em;
437 }
438 nav.toc ul {
439 margin: 0 0.5em 0 0;
440 padding: 0;
441 list-style: none;
442 }
443 nav.toc li {
444 line-height: 1.3em;
445 margin: 0.75em 0;
446 padding-left: 1.2em;
447 text-indent: -1.2em;
448 }
449 /* references */
450 .references dt {
451 text-align: right;
452 font-weight: bold;
453 min-width: 7em;
454 }
455 .references dd {
456 margin-left: 8em;
457 overflow: auto;
458 }
459
460 .refInstance {
461 margin-bottom: 1.25em;
462 }
463
464 .references .ascii {
465 margin-bottom: 0.25em;
466 }
467
468 /* index */
469 .index ul {
470 margin: 0 0 0 1em;
471 padding: 0;
472 list-style: none;
473 }
474 .index ul ul {
475 margin: 0;
476 }
477 .index li {
478 margin: 0;
479 text-indent: -2em;
480 padding-left: 2em;
481 padding-bottom: 5px;
482 }
483 .indexIndex {
484 margin: 0.5em 0 1em;
485 }
486 .index a {
487 font-weight: 700;
488 }
489 /* make the index two-column on all but the smallest screens */
490 @media (min-width: 600px) {
491 .index ul {
492 -moz-column-count: 2;
493 -moz-column-gap: 20px;
494 }
495 .index ul ul {
496 -moz-column-count: 1;
497 -moz-column-gap: 0;
498 }
499 }
500
501 /* authors */
502 address.vcard {
503 font-style: normal;
504 margin: 1em 0;
505 }
506
507 address.vcard .nameRole {
508 font-weight: 700;
509 margin-left: 0;
510 }
511 address.vcard .label {
512 font-family: "Noto Sans",Arial,Helvetica,sans-serif;
513 margin: 0.5em 0;
514 }
515 address.vcard .type {
516 display: none;
517 }
518 .alternative-contact {
519 margin: 1.5em 0 1em;
520 }
521 hr.addr {
522 border-top: 1px dashed;
523 margin: 0;
524 color: #ddd;
525 max-width: calc(100% - 16px);
526 }
527
528 /* temporary notes */
529 .rfcEditorRemove::before {
530 position: absolute;
531 top: 0.2em;
532 right: 0.2em;
533 padding: 0.2em;
534 content: "The RFC Editor will remove this note";
535 color: #9e2a00; /* Arlen: WCAG 2019 */
536 background-color: #ffd; /* Arlen: WCAG 2019 */
537 }
538 .rfcEditorRemove {
539 position: relative;
540 padding-top: 1.8em;
541 background-color: #ffd; /* Arlen: WCAG 2019 */
542 border-radius: 3px;
543 }
544 .cref {
545 background-color: #ffd; /* Arlen: WCAG 2019 */
546 padding: 2px 4px;
547 }
548 .crefSource {
549 font-style: italic;
550 }
551 /* alternative layout for smaller screens */
552 @media screen and (max-width: 1023px) {
553 body {
554 padding-top: 2em;
555 }
556 #title {
557 padding: 1em 0;
558 }
559 h1 {
560 font-size: 24px;
561 }
562 h2 {
563 font-size: 20px;
564 margin-top: -18px; /* provide offset for in-page anchors */
565 padding-top: 38px;
566 }
567 #identifiers dd {
568 max-width: 60%;
569 }
570 #toc {
571 position: fixed;
572 z-index: 2;
573 top: 0;
574 right: 0;
575 padding: 0;
576 margin: 0;
577 background-color: inherit;
578 border-bottom: 1px solid #ccc;
579 }
580 #toc h2 {
581 margin: -1px 0 0 0;
582 padding: 4px 0 4px 6px;
583 padding-right: 1em;
584 min-width: 190px;
585 font-size: 1.1em;
586 text-align: right;
587 background-color: #444;
588 color: white;
589 cursor: pointer;
590 }
591 #toc h2::before { /* css hamburger */
592 float: right;
593 position: relative;
594 width: 1em;
595 height: 1px;
596 left: -164px;
597 margin: 6px 0 0 0;
598 background: white none repeat scroll 0 0;
599 box-shadow: 0 4px 0 0 white, 0 8px 0 0 white;
600 content: "";
601 }
602 #toc nav {
603 display: none;
604 padding: 0.5em 1em 1em;
605 overflow: auto;
606 height: calc(100vh - 48px);
607 border-left: 1px solid #ddd;
608 }
609 }
610
611 /* alternative layout for wide screens */
612 @media screen and (min-width: 1024px) {
613 body {
614 max-width: 724px;
615 margin: 42px auto;
616 padding-left: 1.5em;
617 padding-right: 29em;
618 }
619 #toc {
620 position: fixed;
621 top: 42px;
622 right: 42px;
623 width: 25%;
624 margin: 0;
625 padding: 0 1em;
626 z-index: 1;
627 }
628 #toc h2 {
629 border-top: none;
630 border-bottom: 1px solid #ddd;
631 font-size: 1em;
632 font-weight: normal;
633 margin: 0;
634 padding: 0.25em 1em 1em 0;
635 }
636 #toc nav {
637 display: block;
638 height: calc(90vh - 84px);
639 bottom: 0;
640 padding: 0.5em 0 0;
641 overflow: auto;
642 }
643 img { /* future proofing */
644 max-width: 100%;
645 height: auto;
646 }
647 }
648
649 /* pagination */
650 @media print {
651 body {
652
653 width: 100%;
654 }
655 p {
656 orphans: 3;
657 widows: 3;
658 }
659 #n-copyright-notice {
660 border-bottom: none;
661 }
662 #toc, #n-introduction {
663 page-break-before: always;
664 }
665 #toc {
666 border-top: none;
667 padding-top: 0;
668 }
669 figure, pre {
670 page-break-inside: avoid;
671 }
672 figure {
673 overflow: scroll;
674 }
675 h1, h2, h3, h4, h5, h6 {
676 page-break-after: avoid;
677 }
678 h2+*, h3+*, h4+*, h5+*, h6+* {
679 page-break-before: avoid;
680 }
681 pre {
682 white-space: pre-wrap;
683 word-wrap: break-word;
684 font-size: 10pt;
685 }
686 table {
687 border: 1px solid #ddd;
688 }
689 td {
690 border-top: 1px solid #ddd;
691 }
692 }
693
694 /* This is commented out here, as the string-set: doesn't
695 pass W3C validation currently */
696 /*
697 .ears thead .left {
698 string-set: ears-top-left content();
699 }
700
701 .ears thead .center {
702 string-set: ears-top-center content();
703 }
704
705 .ears thead .right {
706 string-set: ears-top-right content();
707 }
708
709 .ears tfoot .left {
710 string-set: ears-bottom-left content();
711 }
712
713 .ears tfoot .center {
714 string-set: ears-bottom-center content();
715 }
716
717 .ears tfoot .right {
718 string-set: ears-bottom-right content();
719 }
720 */
721
722 @page :first {
723 padding-top: 0;
724 @top-left {
725 content: normal;
726 border: none;
727 }
728 @top-center {
729 content: normal;
730 border: none;
731 }
732 @top-right {
733 content: normal;
734 border: none;
735 }
736 }
737
738 @page {
739 size: A4;
740 margin-bottom: 45mm;
741 padding-top: 20px;
742 /* The follwing is commented out here, but set appropriately by in code, as
743 the content depends on the document */
744 /*
745 @top-left {
746 content: 'Internet-Draft';
747 vertical-align: bottom;
748 border-bottom: solid 1px #ccc;
749 }
750 @top-left {
751 content: string(ears-top-left);
752 vertical-align: bottom;
753 border-bottom: solid 1px #ccc;
754 }
755 @top-center {
756 content: string(ears-top-center);
757 vertical-align: bottom;
758 border-bottom: solid 1px #ccc;
759 }
760 @top-right {
761 content: string(ears-top-right);
762 vertical-align: bottom;
763 border-bottom: solid 1px #ccc;
764 }
765 @bottom-left {
766 content: string(ears-bottom-left);
767 vertical-align: top;
768 border-top: solid 1px #ccc;
769 }
770 @bottom-center {
771 content: string(ears-bottom-center);
772 vertical-align: top;
773 border-top: solid 1px #ccc;
774 }
775 @bottom-right {
776 content: '[Page ' counter(page) ']';
777 vertical-align: top;
778 border-top: solid 1px #ccc;
779 }
780 */
781
782 }
783
784 /* Changes introduced to fix issues found during implementation */
785 /* Make sure links are clickable even if overlapped by following H* */
786 a {
787 z-index: 2;
788 }
789 /* Separate body from document info even without intervening H1 */
790 section {
791 clear: both;
792 }
793
794
795 /* Top align author divs, to avoid names without organization dropping level with org names */
796 .author {
797 vertical-align: top;
798 }
799
800 /* Leave room in document info to show Internet-Draft on one line */
801 #identifiers dt {
802 width: 8em;
803 }
804
805 /* Don't waste quite as much whitespace between label and value in doc info */
806 #identifiers dd {
807 margin-left: 1em;
808 }
809
810 /* Give floating toc a background color (needed when it's a div inside section */
811 #toc {
812 background-color: white;
813 }
814
815 /* Make the collapsed ToC header render white on gray also when it's a link */
816 @media screen and (max-width: 1023px) {
817 #toc h2 a,
818 #toc h2 a:link,
819 #toc h2 a:focus,
820 #toc h2 a:hover,
821 #toc a.toplink,
822 #toc a.toplink:hover {
823 color: white;
824 background-color: #444;
825 text-decoration: none;
826 }
827 }
828
829 /* Give the bottom of the ToC some whitespace */
830 @media screen and (min-width: 1024px) {
831 #toc {
832 padding: 0 0 1em 1em;
833 }
834 }
835
836 /* Style section numbers with more space between number and title */
837 .section-number {
838 padding-right: 0.5em;
839 }
840
841 /* prevent monospace from becoming overly large */
842 tt, code, pre, code {
843 font-size: 95%;
844 }
845
846 /* Fix the height/width aspect for ascii art*/
847 pre.sourcecode,
848 .art-text pre {
849 line-height: 1.12;
850 }
851
852
853 /* Add styling for a link in the ToC that points to the top of the document */
854 a.toplink {
855 float: right;
856 margin-right: 0.5em;
857 }
858
859 /* Fix the dl styling to match the RFC 7992 attributes */
860 dl > dt,
861 dl.dlParallel > dt {
862 float: left;
863 margin-right: 1em;
864 }
865 dl.dlNewline > dt {
866 float: none;
867 }
868
869 /* Provide styling for table cell text alignment */
870 table td.text-left,
871 table th.text-left {
872 text-align: left;
873 }
874 table td.text-center,
875 table th.text-center {
876 text-align: center;
877 }
878 table td.text-right,
879 table th.text-right {
880 text-align: right;
881 }
882
883 /* Make the alternative author contact informatio look less like just another
884 author, and group it closer with the primary author contact information */
885 .alternative-contact {
886 margin: 0.5em 0 0.25em 0;
887 }
888 address .non-ascii {
889 margin: 0 0 0 2em;
890 }
891
892 /* With it being possible to set tables with alignment
893 left, center, and right, { width: 100%; } does not make sense */
894 table {
895 width: auto;
896 }
897
898 /* Avoid reference text that sits in a block with very wide left margin,
899 because of a long floating dt label.*/
900 .references dd {
901 overflow: visible;
902 }
903
904 /* Control caption placement */
905 caption {
906 caption-side: bottom;
907 }
908
909 /* Limit the width of the author address vcard, so names in right-to-left
910 script don't end up on the other side of the page. */
911
912 address.vcard {
913 max-width: 30em;
914 margin-right: auto;
915 }
916
917 /* For address alignment dependent on LTR or RTL scripts */
918 address div.left {
919 text-align: left;
920 }
921 address div.right {
922 text-align: right;
923 }
924
925 /* Provide table alignment support. We can't use the alignX classes above
926 since they do unwanted things with caption and other styling. */
927 table.right {
928 margin-left: auto;
929 margin-right: 0;
930 }
931 table.center {
932 margin-left: auto;
933 margin-right: auto;
934 }
935 table.left {
936 margin-left: 0;
937 margin-right: auto;
938 }
939
940 /* Give the table caption label the same styling as the figcaption */
941 caption a[href] {
942 color: #222;
943 }
944
945 @media print {
946 .toplink {
947 display: none;
948 }
949
950 /* avoid overwriting the top border line with the ToC header */
951 #toc {
952 padding-top: 1px;
953 }
954
955 /* Avoid page breaks inside dl and author address entries */
956 .vcard {
957 page-break-inside: avoid;
958 }
959
960 }
961 /* Tweak the bcp14 keyword presentation */
962 .bcp14 {
963 font-variant: small-caps;
964 font-weight: bold;
965 font-size: 0.9em;
966 }
967 /* Tweak the invisible space above H* in order not to overlay links in text above */
968 h2 {
969 margin-top: -18px; /* provide offset for in-page anchors */
970 padding-top: 31px;
971 }
972 h3 {
973 margin-top: -18px; /* provide offset for in-page anchors */
974 padding-top: 24px;
975 }
976 h4 {
977 margin-top: -18px; /* provide offset for in-page anchors */
978 padding-top: 24px;
979 }
980 /* Float artwork pilcrow to the right */
981 @media screen {
982 .artwork a.pilcrow {
983 display: block;
984 line-height: 0.7;
985 margin-top: 0.15em;
986 }
987 }
988 /* Make pilcrows on dd visible */
989 @media screen {
990 dd:hover > a.pilcrow {
991 visibility: visible;
992 }
993 }
994 /* Make the placement of figcaption match that of a table's caption
995 by removing the figure's added bottom margin */
996 .alignLeft.art-text,
997 .alignCenter.art-text,
998 .alignRight.art-text {
999 margin-bottom: 0;
1000 }
1001 .alignLeft,
1002 .alignCenter,
1003 .alignRight {
1004 margin: 1em 0 0 0;
1005 }
1006 /* In print, the pilcrow won't show on hover, so prevent it from taking up space,
1007 possibly even requiring a new line */
1008 @media print {
1009 a.pilcrow {
1010 display: none;
1011 }
1012 }
1013 /* Styling for the external metadata */
1014 div#external-metadata {
1015 background-color: #eee;
1016 padding: 0.5em;
1017 margin-bottom: 0.5em;
1018 display: none;
1019 }
1020 div#internal-metadata {
1021 padding: 0.5em; /* to match the external-metadata padding */
1022 }
1023 /* Styling for title RFC Number */
1024 h1#rfcnum {
1025 clear: both;
1026 margin: 0 0 -1em;
1027 padding: 1em 0 0 0;
1028 }
1029 /* Make .olPercent look the same as <ol><li> */
1030 dl.olPercent > dd {
1031 margin-bottom: 0.25em;
1032 min-height: initial;
1033 }
1034 /* Give aside some styling to set it apart */
1035 aside {
1036 border-left: 1px solid #ddd;
1037 margin: 1em 0 1em 2em;
1038 padding: 0.2em 2em;
1039 }
1040 aside > dl,
1041 aside > ol,
1042 aside > ul,
1043 aside > table,
1044 aside > p {
1045 margin-bottom: 0.5em;
1046 }
1047 /* Additional page break settings */
1048 @media print {
1049 figcaption, table caption {
1050 page-break-before: avoid;
1051 }
1052 }
1053 /* Font size adjustments for print */
1054 @media print {
1055 body { font-size: 10pt; line-height: normal; max-width: 96%; }
1056 h1 { font-size: 1.72em; padding-top: 1.5em; } /* 1*1.2*1.2*1.2 */
1057 h2 { font-size: 1.44em; padding-top: 1.5em; } /* 1*1.2*1.2 */
1058 h3 { font-size: 1.2em; padding-top: 1.5em; } /* 1*1.2 */
1059 h4 { font-size: 1em; padding-top: 1.5em; }
1060 h5, h6 { font-size: 1em; margin: initial; padding: 0.5em 0 0.3em; }
1061 }
1062 /* Sourcecode margin in print, when there's no pilcrow */
1063 @media print {
1064 .artwork,
1065 .sourcecode {
1066 margin-bottom: 1em;
1067 }
1068 }
1069 /* Avoid narrow tables forcing too narrow table captions, which may render badly */
1070 table {
1071 min-width: 20em;
1072 }
1073 /* ol type a */
1074 ol.type-a { list-style-type: lower-alpha; }
1075 ol.type-A { list-style-type: upper-alpha; }
1076 ol.type-i { list-style-type: lower-roman; }
1077 ol.type-I { list-style-type: lower-roman; }
1078 /* Apply the print table and row borders in general, on request from the RPC,
1079 and increase the contrast between border and odd row background sligthtly */
1080 table {
1081 border: 1px solid #ddd;
1082 }
1083 td {
1084 border-top: 1px solid #ddd;
1085 }
1086 tr:nth-child(2n+1) > td {
1087 background-color: #f8f8f8;
1088 }
1089 /* Use style rules to govern display of the TOC. */
1090 @media screen and (max-width: 1023px) {
1091 #toc nav { display: none; }
1092 #toc.active nav { display: block; }
1093 }
1094 /* Add support for keepWithNext */
1095 .keepWithNext {
1096 break-after: avoid-page;
1097 break-after: avoid-page;
1098 }
1099 /* Add support for keepWithPrevious */
1100 .keepWithPrevious {
1101 break-before: avoid-page;
1102 }
1103 /* Change the approach to avoiding breaks inside artwork etc. */
1104 figure, pre, table, .artwork, .sourcecode {
1105 break-before: auto;
1106 break-after: auto;
1107 }
1108 /* Avoid breaks between <dt> and <dd> */
1109 dl {
1110 break-before: auto;
1111 break-inside: auto;
1112 }
1113 dt {
1114 break-before: auto;
1115 break-after: avoid-page;
1116 }
1117 dd {
1118 break-before: avoid-page;
1119 break-after: auto;
1120 orphans: 3;
1121 widows: 3
1122 }
1123 span.break, dd.break {
1124 margin-bottom: 0;
1125 min-height: 0;
1126 break-before: auto;
1127 break-inside: auto;
1128 break-after: auto;
1129 }
1130 /* Undo break-before ToC */
1131 @media print {
1132 #toc {
1133 break-before: auto;
1134 }
1135 }
1136 /* Text in compact lists should not get extra bottim margin space,
1137 since that would makes the list not compact */
1138 ul.compact p, .ulCompact p,
1139 ol.compact p, .olCompact p {
1140 margin: 0;
1141 }
1142 /* But the list as a whole needs the extra space at the end */
1143 section ul.compact,
1144 section .ulCompact,
1145 section ol.compact,
1146 section .olCompact {
1147 margin-bottom: 1em; /* same as p not within ul.compact etc. */
1148 }
1149 /* The tt and code background above interferes with for instance table cell
1150 backgrounds. Changed to something a bit more selective. */
1151 tt, code {
1152 background-color: transparent;
1153 }
1154 p tt, p code, li tt, li code {
1155 background-color: #f8f8f8;
1156 }
1157 /* Tweak the pre margin -- 0px doesn't come out well */
1158 pre {
1159 margin-top: 0.5px;
1160 }
1161 /* Tweak the comact list text */
1162 ul.compact, .ulCompact,
1163 ol.compact, .olCompact,
1164 dl.compact, .dlCompact {
1165 line-height: normal;
1166 }
1167 /* Don't add top margin for nested lists */
1168 li > ul, li > ol, li > dl,
1169 dd > ul, dd > ol, dd > dl,
1170 dl > dd > dl {
1171 margin-top: initial;
1172 }
1173 /* Elements that should not be rendered on the same line as a <dt> */
1174 /* This should match the element list in writer.text.TextWriter.render_dl() */
1175 dd > div.artwork:first-child,
1176 dd > aside:first-child,
1177 dd > figure:first-child,
1178 dd > ol:first-child,
1179 dd > div:first-child > pre.sourcecode,
1180 dd > table:first-child,
1181 dd > ul:first-child {
1182 clear: left;
1183 }
1184 /* fix for weird browser behaviour when <dd/> is empty */
1185 dt+dd:empty::before{
1186 content: "\00a0";
1187 }
1188 /* Make paragraph spacing inside <li> smaller than in body text, to fit better within the list */
1189 li > p {
1190 margin-bottom: 0.5em
1191 }
1192 /* Don't let p margin spill out from inside list items */
1193 li > p:last-of-type {
1194 margin-bottom: 0;
1195 }
1196 </style>
1197 <link href="rfc-local.css" rel="stylesheet" type="text/css">
1198 <link href="https://dx.doi.org/10.17487/rfc9135" rel="alternate">
1199 <link href="urn:issn:2070-1721" rel="alternate">
1200 <link href="https://datatracker.ietf.org/doc/draft-ietf-bess-evpn-inter-subnet-forwarding-15" rel="prev">
1201 </head>
1202 <body>
1203 <script src="https://www.rfc-editor.org/js/metadata.min.js"></script>
1204 <table class="ears">
1205 <thead><tr>
1206 <td class="left">RFC 9135</td>
1207 <td class="center">IRB EVPN</td>
1208 <td class="right">October 2021</td>
1209 </tr></thead>
1210 <tfoot><tr>
1211 <td class="left">Sajassi, et al.</td>
1212 <td class="center">Standards Track</td>
1213 <td class="right">[Page]</td>
1214 </tr></tfoot>
1215 </table>
1216 <div id="external-metadata" class="document-information"></div>
1217 <div id="internal-metadata" class="document-information">
1218 <dl id="identifiers">
1219 <dt class="label-stream">Stream:</dt>
1220 <dd class="stream">Internet Engineering Task Force (IETF)</dd>
1221 <dt class="label-rfc">RFC:</dt>
1222 <dd class="rfc"><a href="https://www.rfc-editor.org/rfc/rfc9135" class="eref">9135</a></dd>
1223 <dt class="label-category">Category:</dt>
1224 <dd class="category">Standards Track</dd>
1225 <dt class="label-published">Published:</dt>
1226 <dd class="published">
1227 <time datetime="2021-10" class="published">October 2021</time>
1228 </dd>
1229 <dt class="label-issn">ISSN:</dt>
1230 <dd class="issn">2070-1721</dd>
1231 <dt class="label-authors">Authors:</dt>
1232 <dd class="authors">
1233 <div class="author">
1234 <div class="author-name">A. Sajassi</div>
1235 <div class="org">Cisco Systems</div>
1236 </div>
1237 <div class="author">
1238 <div class="author-name">S. Salam</div>
1239 <div class="org">Cisco Systems</div>
1240 </div>
1241 <div class="author">
1242 <div class="author-name">S. Thoria</div>
1243 <div class="org">Cisco Systems</div>
1244 </div>
1245 <div class="author">
1246 <div class="author-name">J. Drake</div>
1247 <div class="org">Juniper</div>
1248 </div>
1249 <div class="author">
1250 <div class="author-name">J. Rabadan</div>
1251 <div class="org">Nokia</div>
1252 </div>
1253 </dd>
1254 </dl>
1255 </div>
1256 <h1 id="rfcnum">RFC 9135</h1>
1257 <h1 id="title">Integrated Routing and Bridging in Ethernet VPN (EVPN)</h1>
1258 <section id="section-abstract">
1259 <h2 id="abstract"><a href="#abstract" class="selfRef">Abstract</a></h2>
1260 <p id="section-abstract-1">
1261 Ethernet VPN (EVPN) provides an extensible and flexible multihoming
1262 VPN solution over an MPLS/IP network for intra-subnet connectivity
1263 among Tenant Systems and end devices that can be physical or virtual.
1264 However, there are scenarios for which there is a need for a dynamic
1265 and efficient inter-subnet connectivity among these Tenant Systems
1266 and end devices while maintaining the multihoming capabilities of
1267 EVPN. This document describes an Integrated Routing and Bridging
1268 (IRB) solution based on EVPN to address such requirements.<a href="#section-abstract-1" class="pilcrow">¶</a></p>
1269 </section>
1270 <div id="status-of-memo">
1271 <section id="section-boilerplate.1">
1272 <h2 id="name-status-of-this-memo">
1273 <a href="#name-status-of-this-memo" class="section-name selfRef">Status of This Memo</a>
1274 </h2>
1275 <p id="section-boilerplate.1-1">
1276 This is an Internet Standards Track document.<a href="#section-boilerplate.1-1" class="pilcrow">¶</a></p>
1277 <p id="section-boilerplate.1-2">
1278 This document is a product of the Internet Engineering Task Force
1279 (IETF). It represents the consensus of the IETF community. It has
1280 received public review and has been approved for publication by
1281 the Internet Engineering Steering Group (IESG). Further
1282 information on Internet Standards is available in Section 2 of
1283 RFC 7841.<a href="#section-boilerplate.1-2" class="pilcrow">¶</a></p>
1284 <p id="section-boilerplate.1-3">
1285 Information about the current status of this document, any
1286 errata, and how to provide feedback on it may be obtained at
1287 <span><a href="https://www.rfc-editor.org/info/rfc9135">https://www.rfc-editor.org/info/rfc9135</a></span>.<a href="#section-boilerplate.1-3" class="pilcrow">¶</a></p>
1288 </section>
1289 </div>
1290 <div id="copyright">
1291 <section id="section-boilerplate.2">
1292 <h2 id="name-copyright-notice">
1293 <a href="#name-copyright-notice" class="section-name selfRef">Copyright Notice</a>
1294 </h2>
1295 <p id="section-boilerplate.2-1">
1296 Copyright (c) 2021 IETF Trust and the persons identified as the
1297 document authors. All rights reserved.<a href="#section-boilerplate.2-1" class="pilcrow">¶</a></p>
1298 <p id="section-boilerplate.2-2">
1299 This document is subject to BCP 78 and the IETF Trust's Legal
1300 Provisions Relating to IETF Documents
1301 (<span><a href="https://trustee.ietf.org/license-info">https://trustee.ietf.org/license-info</a></span>) in effect on the date of
1302 publication of this document. Please review these documents
1303 carefully, as they describe your rights and restrictions with
1304 respect to this document. Code Components extracted from this
1305 document must include Simplified BSD License text as described in
1306 Section 4.e of the Trust Legal Provisions and are provided without
1307 warranty as described in the Simplified BSD License.<a href="#section-boilerplate.2-2" class="pilcrow">¶</a></p>
1308 </section>
1309 </div>
1310 <div id="toc">
1311 <section id="section-toc.1">
1312 <a href="#" onclick="scroll(0,0)" class="toplink">▲</a><h2 id="name-table-of-contents">
1313 <a href="#name-table-of-contents" class="section-name selfRef">Table of Contents</a>
1314 </h2>
1315 <nav class="toc"><ul class="compact toc ulBare ulEmpty">
1316 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.1">
1317 <p id="section-toc.1-1.1.1" class="keepWithNext"><a href="#section-1" class="xref">1</a>.  <a href="#name-introduction" class="xref">Introduction</a></p>
1318 </li>
1319 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.2">
1320 <p id="section-toc.1-1.2.1" class="keepWithNext"><a href="#section-2" class="xref">2</a>.  <a href="#name-terminology" class="xref">Terminology</a></p>
1321 <ul class="compact toc ulBare ulEmpty">
1322 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.2.2.1">
1323 <p id="section-toc.1-1.2.2.1.1" class="keepWithNext"><a href="#section-2.1" class="xref">2.1</a>.  <a href="#name-requirements-language" class="xref">Requirements Language</a></p>
1324 </li>
1325 </ul>
1326 </li>
1327 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.3">
1328 <p id="section-toc.1-1.3.1"><a href="#section-3" class="xref">3</a>.  <a href="#name-evpn-pe-model-for-irb-opera" class="xref">EVPN PE Model for IRB Operation</a></p>
1329 </li>
1330 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.4">
1331 <p id="section-toc.1-1.4.1"><a href="#section-4" class="xref">4</a>.  <a href="#name-symmetric-and-asymmetric-ir" class="xref">Symmetric and Asymmetric IRB</a></p>
1332 <ul class="compact toc ulBare ulEmpty">
1333 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.4.2.1">
1334 <p id="section-toc.1-1.4.2.1.1"><a href="#section-4.1" class="xref">4.1</a>.  <a href="#name-irb-interface-and-its-mac-a" class="xref">IRB Interface and Its MAC and IP Addresses</a></p>
1335 </li>
1336 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.4.2.2">
1337 <p id="section-toc.1-1.4.2.2.1"><a href="#section-4.2" class="xref">4.2</a>.  <a href="#name-operational-considerations" class="xref">Operational Considerations</a></p>
1338 </li>
1339 </ul>
1340 </li>
1341 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.5">
1342 <p id="section-toc.1-1.5.1"><a href="#section-5" class="xref">5</a>.  <a href="#name-symmetric-irb-procedures" class="xref">Symmetric IRB Procedures</a></p>
1343 <ul class="compact toc ulBare ulEmpty">
1344 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.5.2.1">
1345 <p id="section-toc.1-1.5.2.1.1"><a href="#section-5.1" class="xref">5.1</a>.  <a href="#name-control-plane-advertising-p" class="xref">Control Plane - Advertising PE</a></p>
1346 </li>
1347 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.5.2.2">
1348 <p id="section-toc.1-1.5.2.2.1"><a href="#section-5.2" class="xref">5.2</a>.  <a href="#name-control-plane-receiving-pe" class="xref">Control Plane - Receiving PE</a></p>
1349 </li>
1350 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.5.2.3">
1351 <p id="section-toc.1-1.5.2.3.1"><a href="#section-5.3" class="xref">5.3</a>.  <a href="#name-subnet-route-advertisement" class="xref">Subnet Route Advertisement</a></p>
1352 </li>
1353 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.5.2.4">
1354 <p id="section-toc.1-1.5.2.4.1"><a href="#section-5.4" class="xref">5.4</a>.  <a href="#name-data-plane-ingress-pe" class="xref">Data Plane - Ingress PE</a></p>
1355 </li>
1356 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.5.2.5">
1357 <p id="section-toc.1-1.5.2.5.1"><a href="#section-5.5" class="xref">5.5</a>.  <a href="#name-data-plane-egress-pe" class="xref">Data Plane - Egress PE</a></p>
1358 </li>
1359 </ul>
1360 </li>
1361 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.6">
1362 <p id="section-toc.1-1.6.1"><a href="#section-6" class="xref">6</a>.  <a href="#name-asymmetric-irb-procedures" class="xref">Asymmetric IRB Procedures</a></p>
1363 <ul class="compact toc ulBare ulEmpty">
1364 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.6.2.1">
1365 <p id="section-toc.1-1.6.2.1.1"><a href="#section-6.1" class="xref">6.1</a>.  <a href="#name-control-plane-advertising-pe" class="xref">Control Plane - Advertising PE</a></p>
1366 </li>
1367 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.6.2.2">
1368 <p id="section-toc.1-1.6.2.2.1"><a href="#section-6.2" class="xref">6.2</a>.  <a href="#name-control-plane-receiving-pe-2" class="xref">Control Plane - Receiving PE</a></p>
1369 </li>
1370 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.6.2.3">
1371 <p id="section-toc.1-1.6.2.3.1"><a href="#section-6.3" class="xref">6.3</a>.  <a href="#name-data-plane-ingress-pe-2" class="xref">Data Plane - Ingress PE</a></p>
1372 </li>
1373 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.6.2.4">
1374 <p id="section-toc.1-1.6.2.4.1"><a href="#section-6.4" class="xref">6.4</a>.  <a href="#name-data-plane-egress-pe-2" class="xref">Data Plane - Egress PE</a></p>
1375 </li>
1376 </ul>
1377 </li>
1378 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.7">
1379 <p id="section-toc.1-1.7.1"><a href="#section-7" class="xref">7</a>.  <a href="#name-mobility-procedure" class="xref">Mobility Procedure</a></p>
1380 <ul class="compact toc ulBare ulEmpty">
1381 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.7.2.1">
1382 <p id="section-toc.1-1.7.2.1.1"><a href="#section-7.1" class="xref">7.1</a>.  <a href="#name-initiating-a-gratuitous-arp" class="xref">Initiating a Gratuitous ARP upon a Move</a></p>
1383 </li>
1384 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.7.2.2">
1385 <p id="section-toc.1-1.7.2.2.1"><a href="#section-7.2" class="xref">7.2</a>.  <a href="#name-sending-data-traffic-withou" class="xref">Sending Data Traffic without an ARP Request</a></p>
1386 </li>
1387 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.7.2.3">
1388 <p id="section-toc.1-1.7.2.3.1"><a href="#section-7.3" class="xref">7.3</a>.  <a href="#name-silent-host" class="xref">Silent Host</a></p>
1389 </li>
1390 </ul>
1391 </li>
1392 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.8">
1393 <p id="section-toc.1-1.8.1"><a href="#section-8" class="xref">8</a>.  <a href="#name-bgp-encoding" class="xref">BGP Encoding</a></p>
1394 <ul class="compact toc ulBare ulEmpty">
1395 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.8.2.1">
1396 <p id="section-toc.1-1.8.2.1.1"><a href="#section-8.1" class="xref">8.1</a>.  <a href="#name-evpn-routers-mac-extended-c" class="xref">EVPN Router's MAC Extended Community</a></p>
1397 </li>
1398 </ul>
1399 </li>
1400 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.9">
1401 <p id="section-toc.1-1.9.1"><a href="#section-9" class="xref">9</a>.  <a href="#name-operational-models-for-symm" class="xref">Operational Models for Symmetric Inter-Subnet Forwarding</a></p>
1402 <ul class="compact toc ulBare ulEmpty">
1403 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.9.2.1">
1404 <p id="section-toc.1-1.9.2.1.1"><a href="#section-9.1" class="xref">9.1</a>.  <a href="#name-irb-forwarding-on-nves-for-" class="xref">IRB Forwarding on NVEs for Tenant Systems</a></p>
1405 <ul class="compact toc ulBare ulEmpty">
1406 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.9.2.1.2.1">
1407 <p id="section-toc.1-1.9.2.1.2.1.1"><a href="#section-9.1.1" class="xref">9.1.1</a>.  <a href="#name-control-plane-operation" class="xref">Control Plane Operation</a></p>
1408 </li>
1409 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.9.2.1.2.2">
1410 <p id="section-toc.1-1.9.2.1.2.2.1"><a href="#section-9.1.2" class="xref">9.1.2</a>.  <a href="#name-data-plane-operation" class="xref">Data Plane Operation</a></p>
1411 </li>
1412 </ul>
1413 </li>
1414 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.9.2.2">
1415 <p id="section-toc.1-1.9.2.2.1"><a href="#section-9.2" class="xref">9.2</a>.  <a href="#name-irb-forwarding-on-nves-for-s" class="xref">IRB Forwarding on NVEs for Subnets behind Tenant Systems</a></p>
1416 <ul class="compact toc ulBare ulEmpty">
1417 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.9.2.2.2.1">
1418 <p id="section-toc.1-1.9.2.2.2.1.1"><a href="#section-9.2.1" class="xref">9.2.1</a>.  <a href="#name-control-plane-operation-2" class="xref">Control Plane Operation</a></p>
1419 </li>
1420 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.9.2.2.2.2">
1421 <p id="section-toc.1-1.9.2.2.2.2.1"><a href="#section-9.2.2" class="xref">9.2.2</a>.  <a href="#name-data-plane-operation-2" class="xref">Data Plane Operation</a></p>
1422 </li>
1423 </ul>
1424 </li>
1425 </ul>
1426 </li>
1427 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.10">
1428 <p id="section-toc.1-1.10.1"><a href="#section-10" class="xref">10</a>. <a href="#name-security-considerations" class="xref">Security Considerations</a></p>
1429 </li>
1430 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.11">
1431 <p id="section-toc.1-1.11.1"><a href="#section-11" class="xref">11</a>. <a href="#name-iana-considerations" class="xref">IANA Considerations</a></p>
1432 </li>
1433 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.12">
1434 <p id="section-toc.1-1.12.1"><a href="#section-12" class="xref">12</a>. <a href="#name-references" class="xref">References</a></p>
1435 <ul class="compact toc ulBare ulEmpty">
1436 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.12.2.1">
1437 <p id="section-toc.1-1.12.2.1.1"><a href="#section-12.1" class="xref">12.1</a>.  <a href="#name-normative-references" class="xref">Normative References</a></p>
1438 </li>
1439 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.12.2.2">
1440 <p id="section-toc.1-1.12.2.2.1"><a href="#section-12.2" class="xref">12.2</a>.  <a href="#name-informative-references" class="xref">Informative References</a></p>
1441 </li>
1442 </ul>
1443 </li>
1444 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.13">
1445 <p id="section-toc.1-1.13.1"><a href="#appendix-A" class="xref"></a><a href="#name-acknowledgements" class="xref">Acknowledgements</a></p>
1446 </li>
1447 <li class="compact toc ulBare ulEmpty" id="section-toc.1-1.14">
1448 <p id="section-toc.1-1.14.1"><a href="#appendix-B" class="xref"></a><a href="#name-authors-addresses" class="xref">Authors' Addresses</a></p>
1449 </li>
1450 </ul>
1451 </nav>
1452 </section>
1453 </div>
1454 <div id="intro">
1455 <section id="section-1">
1456 <h2 id="name-introduction">
1457 <a href="#section-1" class="section-number selfRef">1. </a><a href="#name-introduction" class="section-name selfRef">Introduction</a>
1458 </h2>
1459 <p id="section-1-1">
1460 EVPN <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span> provides an extensible and flexible multihoming VPN
1461 solution over an MPLS/IP network for intra-subnet connectivity among
1462 Tenant Systems (TSs) and end devices that can be physical or
1463 virtual, where an IP subnet is represented by an EVPN instance (EVI)
1464 for a VLAN-based service or by an (EVI, VLAN) association for a VLAN-aware bundle
1465 service. However, there are scenarios for which there is a need for
1466 a dynamic and efficient inter-subnet connectivity among these Tenant
1467 Systems and end devices while maintaining the multihoming
1468 capabilities of EVPN. This document describes an Integrated Routing
1469 and Bridging (IRB) solution based on EVPN to address such
1470 requirements.<a href="#section-1-1" class="pilcrow">¶</a></p>
1471 <p id="section-1-2">
1472 Inter-subnet communication is typically performed by centralized Layer 3 (L3) gateway (GW) devices, which enforce all inter-subnet communication policies
1473 and perform all inter-subnet forwarding. When two TSs belonging to two different
1474 subnets connected to the same Provider Edge (PE) wanted to communicate with each
1475 other, their traffic needed to be backhauled from the PE all the way
1476 to the centralized gateway where inter-subnet switching is performed
1477 and then sent back to the PE. For today's large multi-tenant Data Center (DC),
1478 this scheme is very inefficient and sometimes impractical.<a href="#section-1-2" class="pilcrow">¶</a></p>
1479 <p id="section-1-3">
1480 In order to overcome the drawback of the centralized L3 GW
1481 approach, IRB functionality is needed on the PEs (also referred to as
1482 EVPN Network Virtualization Edges (NVEs)) attached to TSs in order to avoid inefficient forwarding
1483 of tenant traffic (i.e., avoid backhauling and hair pinning). When
1484 a PE with IRB capability receives tenant traffic over an Attachment
1485 Circuit (AC), it cannot only locally bridge the tenant intra-subnet
1486 traffic but also locally route the tenant inter-subnet traffic on
1487 a packet-by-packet basis, thus meeting the requirements for both intra-
1488 and inter-subnet forwarding and avoiding non-optimal traffic
1489 forwarding associated with a centralized L3 GW approach.<a href="#section-1-3" class="pilcrow">¶</a></p>
1490 <p id="section-1-4">
1491 Some TSs run non-IP protocols in conjunction with their IP traffic.
1492 Therefore, it is important to handle both kinds of traffic optimally --
1493 e.g., to bridge non-IP and intra-subnet traffic and to route inter-subnet
1494 IP traffic. Therefore, the solution needs to meet the following
1495 requirements:<a href="#section-1-4" class="pilcrow">¶</a></p>
1496 <span class="break"></span><dl class="dlParallel" id="section-1-5">
1497 <dt id="section-1-5.1">R1:</dt>
1498 <dd style="margin-left: 1.5em" id="section-1-5.2"> The solution must provide each tenant with IP routing of its
1499 inter-subnet traffic and Ethernet bridging of its intra-subnet
1500 traffic and non-routable traffic, where non-routable traffic refers
1501 to both non-IP traffic and IP traffic whose version differs from the
1502 IP version configured in IP Virtual Routing and Forwarding (IP-VRF). For example, if an IP-VRF in an
1503 NVE is configured for IPv6 and that NVE receives IPv4 traffic on the
1504 corresponding VLAN, then the IPv4 traffic is treated as non-routable
1505 traffic.<a href="#section-1-5.2" class="pilcrow">¶</a>
1506 </dd>
1507 <dd class="break"></dd>
1508 <dt id="section-1-5.3">
1509 R2:</dt>
1510 <dd style="margin-left: 1.5em" id="section-1-5.4"> The solution must allow IP routing of inter-subnet traffic to be
1511 disabled on a per-VLAN basis on those PEs that are backhauling that
1512 traffic to another PE for routing.<a href="#section-1-5.4" class="pilcrow">¶</a>
1513 </dd>
1514 <dd class="break"></dd>
1515 </dl>
1516 </section>
1517 </div>
1518 <div id="terms">
1519 <section id="section-2">
1520 <h2 id="name-terminology">
1521 <a href="#section-2" class="section-number selfRef">2. </a><a href="#name-terminology" class="section-name selfRef">Terminology</a>
1522 </h2>
1523 <span class="break"></span><dl class="dlParallel" id="section-2-1">
1524 <dt id="section-2-1.1">AC:</dt>
1525 <dd style="margin-left: 5.0em" id="section-2-1.2">Attachment Circuit<a href="#section-2-1.2" class="pilcrow">¶</a>
1526 </dd>
1527 <dd class="break"></dd>
1528 <dt id="section-2-1.3">ARP:</dt>
1529 <dd style="margin-left: 5.0em" id="section-2-1.4">Address Resolution Protocol<a href="#section-2-1.4" class="pilcrow">¶</a>
1530 </dd>
1531 <dd class="break"></dd>
1532 <dt id="section-2-1.5">ARP Table:</dt>
1533 <dd style="margin-left: 5.0em" id="section-2-1.6"> A logical view of a forwarding table on a PE that
1534 maintains an IP to a MAC binding entry on an IP interface for both IPv4
1535 and IPv6. These entries are learned through ARP/ND or through EVPN.<a href="#section-2-1.6" class="pilcrow">¶</a>
1536 </dd>
1537 <dd class="break"></dd>
1538 <dt id="section-2-1.7">BD:</dt>
1539 <dd style="margin-left: 5.0em" id="section-2-1.8">Broadcast Domain. As per <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span>, an EVI consists of a single BD or multiple
1540 BDs. In the case of VLAN-bundle and VLAN-based service
1541 models (see <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span>), a BD is equivalent to an EVI. In the
1542 case of a VLAN-aware bundle service model, an EVI contains multiple BDs. Also, in this document, "BD" and "subnet" are
1543 equivalent terms, and wherever "subnet" is used, it means "IP subnet".<a href="#section-2-1.8" class="pilcrow">¶</a>
1544 </dd>
1545 <dd class="break"></dd>
1546 <dt id="section-2-1.9">BD Route Target:</dt>
1547 <dd style="margin-left: 5.0em" id="section-2-1.10">Refers to the broadcast-domain-assigned Route Target <span>[<a href="#RFC4364" class="xref">RFC4364</a>]</span>. In the case of a VLAN-aware bundle
1548 service model, all the BD instances in the MAC-VRF
1549 share the same Route Target.<a href="#section-2-1.10" class="pilcrow">¶</a>
1550 </dd>
1551 <dd class="break"></dd>
1552 <dt id="section-2-1.11">BT:</dt>
1553 <dd style="margin-left: 5.0em" id="section-2-1.12">Bridge Table. The instantiation of a BD in a MAC-VRF,
1554 as per <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span>.<a href="#section-2-1.12" class="pilcrow">¶</a>
1555 </dd>
1556 <dd class="break"></dd>
1557 <dt id="section-2-1.13">CE:</dt>
1558 <dd style="margin-left: 5.0em" id="section-2-1.14">Customer Edge<a href="#section-2-1.14" class="pilcrow">¶</a>
1559 </dd>
1560 <dd class="break"></dd>
1561 <dt id="section-2-1.15">DA:</dt>
1562 <dd style="margin-left: 5.0em" id="section-2-1.16">Destination Address<a href="#section-2-1.16" class="pilcrow">¶</a>
1563 </dd>
1564 <dd class="break"></dd>
1565 <dt id="section-2-1.17">Ethernet NVO Tunnel:</dt>
1566 <dd style="margin-left: 5.0em" id="section-2-1.18">Refers to Network Virtualization Overlay tunnels
1567 with an Ethernet payload, as specified for VXLAN in <span>[<a href="#RFC7348" class="xref">RFC7348</a>]</span> and for
1568 NVGRE in <span>[<a href="#RFC7637" class="xref">RFC7637</a>]</span>.<a href="#section-2-1.18" class="pilcrow">¶</a>
1569 </dd>
1570 <dd class="break"></dd>
1571 <dt id="section-2-1.19">EVI:</dt>
1572 <dd style="margin-left: 5.0em" id="section-2-1.20">EVPN Instance spanning NVE/PE devices that are participating
1573 on that EVPN, as per <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span>.<a href="#section-2-1.20" class="pilcrow">¶</a>
1574 </dd>
1575 <dd class="break"></dd>
1576 <dt id="section-2-1.21">EVPN:</dt>
1577 <dd style="margin-left: 5.0em" id="section-2-1.22">Ethernet VPN, as per <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span>.<a href="#section-2-1.22" class="pilcrow">¶</a>
1578 </dd>
1579 <dd class="break"></dd>
1580 <dt id="section-2-1.23">IP NVO Tunnel:</dt>
1581 <dd style="margin-left: 5.0em" id="section-2-1.24">Refers to Network Virtualization Overlay tunnels
1582 with IP payload (no MAC header in the payload) as specified for Generic Protocol Extension (GPE)
1583 in <span>[<a href="#I-D.ietf-nvo3-vxlan-gpe" class="xref">VXLAN-GPE</a>]</span>.<a href="#section-2-1.24" class="pilcrow">¶</a>
1584 </dd>
1585 <dd class="break"></dd>
1586 <dt id="section-2-1.25">IP-VRF:</dt>
1587 <dd style="margin-left: 5.0em" id="section-2-1.26">A Virtual Routing and Forwarding table for IP routes on an
1588 NVE/PE. The IP routes could be populated by EVPN and IP-VPN address
1589 families. An IP-VRF is also an instantiation of a Layer 3 VPN in an
1590 NVE/PE.<a href="#section-2-1.26" class="pilcrow">¶</a>
1591 </dd>
1592 <dd class="break"></dd>
1593 <dt id="section-2-1.27">IRB:</dt>
1594 <dd style="margin-left: 5.0em" id="section-2-1.28">Integrated Routing and Bridging interface. It connects an IP-VRF to a
1595 BD (or subnet).<a href="#section-2-1.28" class="pilcrow">¶</a>
1596 </dd>
1597 <dd class="break"></dd>
1598 <dt id="section-2-1.29">MAC:</dt>
1599 <dd style="margin-left: 5.0em" id="section-2-1.30">Media Access Control<a href="#section-2-1.30" class="pilcrow">¶</a>
1600 </dd>
1601 <dd class="break"></dd>
1602 <dt id="section-2-1.31">MAC-VRF:</dt>
1603 <dd style="margin-left: 5.0em" id="section-2-1.32">A Virtual Routing and Forwarding table for
1604 MAC addresses on an NVE/PE, as per <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span>. A MAC-VRF is
1605 also an instantiation of an EVI in an NVE/PE.<a href="#section-2-1.32" class="pilcrow">¶</a>
1606 </dd>
1607 <dd class="break"></dd>
1608 <dt id="section-2-1.33">ND:</dt>
1609 <dd style="margin-left: 5.0em" id="section-2-1.34">Neighbor Discovery<a href="#section-2-1.34" class="pilcrow">¶</a>
1610 </dd>
1611 <dd class="break"></dd>
1612 <dt id="section-2-1.35">NVE:</dt>
1613 <dd style="margin-left: 5.0em" id="section-2-1.36">Network Virtualization Edge<a href="#section-2-1.36" class="pilcrow">¶</a>
1614 </dd>
1615 <dd class="break"></dd>
1616 <dt id="section-2-1.37">NVGRE:</dt>
1617 <dd style="margin-left: 5.0em" id="section-2-1.38">Network Virtualization Using Generic Routing Encapsulation, as per
1618 <span>[<a href="#RFC7637" class="xref">RFC7637</a>]</span>.<a href="#section-2-1.38" class="pilcrow">¶</a>
1619 </dd>
1620 <dd class="break"></dd>
1621 <dt id="section-2-1.39">NVO:</dt>
1622 <dd style="margin-left: 5.0em" id="section-2-1.40">Network Virtualization Overlay<a href="#section-2-1.40" class="pilcrow">¶</a>
1623 </dd>
1624 <dd class="break"></dd>
1625 <dt id="section-2-1.41">PE:</dt>
1626 <dd style="margin-left: 5.0em" id="section-2-1.42">Provider Edge<a href="#section-2-1.42" class="pilcrow">¶</a>
1627 </dd>
1628 <dd class="break"></dd>
1629 <dt id="section-2-1.43">RT-2:</dt>
1630 <dd style="margin-left: 5.0em" id="section-2-1.44">EVPN Route Type 2, i.e., MAC/IP Advertisement route, as defined
1631 in <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span>.<a href="#section-2-1.44" class="pilcrow">¶</a>
1632 </dd>
1633 <dd class="break"></dd>
1634 <dt id="section-2-1.45">RT-5:</dt>
1635 <dd style="margin-left: 5.0em" id="section-2-1.46">EVPN Route Type 5, i.e., IP Prefix route, as defined in <span><a href="https://www.rfc-editor.org/rfc/rfc9136#section-3" class="relref">Section 3</a> of [<a href="#RFC9136" class="xref">RFC9136</a>]</span>.<a href="#section-2-1.46" class="pilcrow">¶</a>
1636 </dd>
1637 <dd class="break"></dd>
1638 <dt id="section-2-1.47">SA:</dt>
1639 <dd style="margin-left: 5.0em" id="section-2-1.48">Source Address<a href="#section-2-1.48" class="pilcrow">¶</a>
1640 </dd>
1641 <dd class="break"></dd>
1642 <dt id="section-2-1.49">TS:</dt>
1643 <dd style="margin-left: 5.0em" id="section-2-1.50">Tenant System<a href="#section-2-1.50" class="pilcrow">¶</a>
1644 </dd>
1645 <dd class="break"></dd>
1646 <dt id="section-2-1.51">VA:</dt>
1647 <dd style="margin-left: 5.0em" id="section-2-1.52">Virtual Appliance<a href="#section-2-1.52" class="pilcrow">¶</a>
1648 </dd>
1649 <dd class="break"></dd>
1650 <dt id="section-2-1.53">VNI:</dt>
1651 <dd style="margin-left: 5.0em" id="section-2-1.54">Virtual Network Identifier. As in <span>[<a href="#RFC8365" class="xref">RFC8365</a>]</span>, the term is used
1652 as a representation of a 24-bit NVO instance identifier, with the
1653 understanding that "VNI" will refer to a VXLAN Network Identifier in
1654 VXLAN, or a Virtual Subnet Identifier in NVGRE, etc., unless it is
1655 stated otherwise.<a href="#section-2-1.54" class="pilcrow">¶</a>
1656 </dd>
1657 <dd class="break"></dd>
1658 <dt id="section-2-1.55">VTEP:</dt>
1659 <dd style="margin-left: 5.0em" id="section-2-1.56">VXLAN Termination End Point, as per <span>[<a href="#RFC7348" class="xref">RFC7348</a>]</span>.<a href="#section-2-1.56" class="pilcrow">¶</a>
1660 </dd>
1661 <dd class="break"></dd>
1662 <dt id="section-2-1.57">VXLAN:</dt>
1663 <dd style="margin-left: 5.0em" id="section-2-1.58">Virtual eXtensible Local Area Network, as per <span>[<a href="#RFC7348" class="xref">RFC7348</a>]</span>.<a href="#section-2-1.58" class="pilcrow">¶</a>
1664 </dd>
1665 <dd class="break"></dd>
1666 </dl>
1667 <p id="section-2-2">
1668 This document also assumes familiarity with the terminology of <span>[<a href="#RFC7365" class="xref">RFC7365</a>]</span>, <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span>, and <span>[<a href="#RFC8365" class="xref">RFC8365</a>]</span>.<a href="#section-2-2" class="pilcrow">¶</a></p>
1669 <div id="sect-1.1">
1670 <section id="section-2.1">
1671 <h3 id="name-requirements-language">
1672 <a href="#section-2.1" class="section-number selfRef">2.1. </a><a href="#name-requirements-language" class="section-name selfRef">Requirements Language</a>
1673 </h3>
1674 <p id="section-2.1-1">
1675 The key words "<span class="bcp14">MUST</span>", "<span class="bcp14">MUST NOT</span>", "<span class="bcp14">REQUIRED</span>", "<span class="bcp14">SHALL</span>", "<span class="bcp14">SHALL NOT</span>", "<span class="bcp14">SHOULD</span>", "<span class="bcp14">SHOULD NOT</span>", "<span class="bcp14">RECOMMENDED</span>", "<span class="bcp14">NOT RECOMMENDED</span>",
1676 "<span class="bcp14">MAY</span>", and "<span class="bcp14">OPTIONAL</span>" in this document are to be interpreted as
1677 described in BCP 14 <span>[<a href="#RFC2119" class="xref">RFC2119</a>]</span> <span>[<a href="#RFC8174" class="xref">RFC8174</a>]</span>
1678 when, and only when, they appear in all capitals, as shown here.<a href="#section-2.1-1" class="pilcrow">¶</a></p>
1679 </section>
1680 </div>
1681 </section>
1682 </div>
1683 <div id="sect-3">
1684 <section id="section-3">
1685 <h2 id="name-evpn-pe-model-for-irb-opera">
1686 <a href="#section-3" class="section-number selfRef">3. </a><a href="#name-evpn-pe-model-for-irb-opera" class="section-name selfRef">EVPN PE Model for IRB Operation</a>
1687 </h2>
1688 <p id="section-3-1">
1689 Since this document discusses IRB operation in relationship to EVPN
1690 MAC-VRF, IP-VRF, EVI, BD, bridge table, and IRB
1691 interfaces, it is important to understand the relationship between
1692 these components. Therefore, the PE model is illustrated
1693 below to a) describe these components and b) illustrate the
1694 relationship among them.<a href="#section-3-1" class="pilcrow">¶</a></p>
1695 <span id="name-evpn-irb-pe-model"></span><div id="fig-1">
1696 <figure id="figure-1">
1697 <div class="alignLeft art-text artwork" id="section-3-2.1">
1698 <pre>
1699 +-------------------------------------------------------------+
1700 | |
1701 | +------------------+ IRB PE |
1702 | Attachment | +------------------+ |
1703 | Circuit(AC1) | | +----------+ | MPLS/NVO tnl
1704 ----------------------*Bridge | | +-----
1705 | | | |Table(BT1)| | +-----------+ / \ \
1706 | | | | *---------* |&lt;--&gt; |Eth|
1707 | | | | VLAN x | |IRB1| | \ / /
1708 | | | +----------+ | | | +-----
1709 | | | ... | | IP-VRF1 | |
1710 | | | +----------+ | | RD2/RT2 |MPLS/NVO tnl
1711 | | | |Bridge | | | | +-----
1712 | | | |Table(BT2)| |IRB2| | / \ \
1713 | | | | *---------* |&lt;--&gt; |IP |
1714 ----------------------* VLAN y | | +-----------+ \ / /
1715 | AC2 | | +----------+ | +-----
1716 | | | MAC-VRF1 | |
1717 | +-+ RD1/RT1 | |
1718 | +------------------+ |
1719 | |
1720 | |
1721 +-------------------------------------------------------------+
1722 </pre>
1723 </div>
1724 <figcaption><a href="#figure-1" class="selfRef">Figure 1</a>:
1725 <a href="#name-evpn-irb-pe-model" class="selfRef">EVPN IRB PE Model</a>
1726 </figcaption></figure>
1727 </div>
1728 <p id="section-3-3">
1729 A tenant needing IRB services on a PE requires an IP-VRF table along with one or more MAC-VRF tables. An IP-VRF, as defined in <span>[<a href="#RFC4364" class="xref">RFC4364</a>]</span>, is the
1730 instantiation of an IP-VPN instance in a PE. A MAC-VRF, as defined in
1731 <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span>, is the instantiation of an EVI in a PE. A
1732 MAC-VRF consists of one or more bridge tables, where each bridge table
1733 corresponds to a VLAN (broadcast domain). If service interfaces for an
1734 EVPN PE are configured in VLAN-based mode (i.e., <span><a href="https://www.rfc-editor.org/rfc/rfc7432#section-6.1" class="relref">Section 6.1</a> of [<a href="#RFC7432" class="xref">RFC7432</a>]</span>),
1735 then there is only a single bridge table per MAC-VRF (per EVI) -- i.e.,
1736 there is only one tenant VLAN per EVI. However, if service interfaces for
1737 an EVPN PE are configured in VLAN-aware bundle mode (i.e., <span><a href="https://www.rfc-editor.org/rfc/rfc7432#section-6.3" class="relref">Section 6.3</a> of [<a href="#RFC7432" class="xref">RFC7432</a>]</span>), then there are several bridge tables per MAC-VRF (per EVI) --
1738 i.e., there are several tenant VLANs per EVI.<a href="#section-3-3" class="pilcrow">¶</a></p>
1739 <p id="section-3-4">
1740 Each bridge table is connected to an IP-VRF via an L3 interface
1741 called an "IRB interface". Since a single tenant subnet is typically (and
1742 in this document) represented by a VLAN (and thus supported by a
1743 single bridge table), for a given tenant, there are as many bridge
1744 tables as there are subnets. Thus, there are also as many IRB
1745 interfaces between the tenant IP-VRF and the associated bridge tables
1746 as shown in the PE model above.<a href="#section-3-4" class="pilcrow">¶</a></p>
1747 <p id="section-3-5">
1748 IP-VRF is identified by its corresponding Route Target and Route
1749 Distinguisher, and MAC-VRF is also identified by its corresponding Route
1750 Target and Route Distinguisher. If operating in EVPN VLAN-based mode, then
1751 a receiving PE that receives an EVPN route with a MAC-VRF Route Target can
1752 identify the corresponding bridge table; however, if operating in EVPN
1753 VLAN-aware bundle mode, then the receiving PE needs both the MAC-VRF Route
1754 Target and VLAN ID in order to identify the corresponding bridge table.<a href="#section-3-5" class="pilcrow">¶</a></p>
1755 </section>
1756 </div>
1757 <div id="sect-4">
1758 <section id="section-4">
1759 <h2 id="name-symmetric-and-asymmetric-ir">
1760 <a href="#section-4" class="section-number selfRef">4. </a><a href="#name-symmetric-and-asymmetric-ir" class="section-name selfRef">Symmetric and Asymmetric IRB</a>
1761 </h2>
1762 <p id="section-4-1">
1763 This document defines and describes two types of IRB solutions --
1764 namely, symmetric and asymmetric IRB. The description of symmetric
1765 and asymmetric IRB procedures relating to data path operations and
1766 tables in this document is a logical view of data path lookups and
1767 related tables. Actual implementations, while following this logical
1768 view, may not strictly adhere to it for performance trade-offs.
1769 Specifically,<a href="#section-4-1" class="pilcrow">¶</a></p>
1770 <ul class="normal">
1771 <li class="normal" id="section-4-2.1">References to an ARP table in the context of asymmetric IRB is a
1772 logical view of a forwarding table that maintains an IP-to-MAC
1773 binding entry on a Layer 3 interface for both IPv4 and IPv6.
1774 These entries are not subject to ARP or ND protocols. For IP-to-MAC bindings learned via EVPN, an implementation may choose to
1775 import these bindings directly to the respective forwarding table
1776 (such as an adjacency/next-hop table) as opposed to importing them
1777 to ARP or ND protocol tables.<a href="#section-4-2.1" class="pilcrow">¶</a>
1778 </li>
1779 <li class="normal" id="section-4-2.2">References to a host IP lookup followed by a host MAC lookup in the
1780 context of asymmetric IRB <span class="bcp14">MAY</span> be collapsed into a single IP lookup
1781 in a hardware implementation.<a href="#section-4-2.2" class="pilcrow">¶</a>
1782 </li>
1783 </ul>
1784 <p id="section-4-3">
1785 In symmetric IRB, as its name implies, the lookup operation is
1786 symmetric at both the ingress and egress PEs -- i.e., both ingress and
1787 egress PEs perform lookups on both MAC and IP addresses. The ingress
1788 PE performs a MAC lookup followed by an IP lookup, and the egress PE
1789 performs an IP lookup followed by a MAC lookup, as depicted in the
1790 following figure.<a href="#section-4-3" class="pilcrow">¶</a></p>
1791 <span id="name-symmetric-irb"></span><div id="fig-2">
1792 <figure id="figure-2">
1793 <div class="alignLeft art-text artwork" id="section-4-4.1">
1794 <pre>
1795 Ingress PE Egress PE
1796 +-------------------+ +------------------+
1797 | | | |
1798 | +-&gt; IP-VRF ----|----&gt;---|-----&gt; IP-VRF -+ |
1799 | | | | | |
1800 | BT1 BT2 | | BT3 BT2 |
1801 | | | | | |
1802 | ^ | | v |
1803 | | | | | |
1804 +-------------------+ +------------------+
1805 ^ |
1806 | |
1807 TS1-&gt;-+ +-&gt;-TS2
1808 </pre>
1809 </div>
1810 <figcaption><a href="#figure-2" class="selfRef">Figure 2</a>:
1811 <a href="#name-symmetric-irb" class="selfRef">Symmetric IRB</a>
1812 </figcaption></figure>
1813 </div>
1814 <p id="section-4-5">
1815 In symmetric IRB, as shown in <a href="#fig-2" class="xref">Figure 2</a>, the inter-subnet forwarding
1816 between two PEs is done between their associated IP-VRFs. Therefore,
1817 the tunnel connecting these IP-VRFs can be either an IP-only tunnel
1818 (e.g., in the case of MPLS or GPE encapsulation) or an Ethernet NVO tunnel
1819 (e.g., in the case of VXLAN encapsulation). If it is an Ethernet NVO
1820 tunnel, the TS1's IP packet is encapsulated in an Ethernet header
1821 consisting of ingress and egress PE MAC addresses -- i.e., there is
1822 no need for the ingress PE to use the destination TS2's MAC address.
1823
1824 Therefore, in symmetric IRB, there is no need for the ingress PE to
1825 maintain ARP entries for the association of the destination TS2's IP and MAC addresses in its ARP table.
1826
1827 Each PE participating in symmetric IRB
1828 only maintains ARP entries for locally connected hosts and
1829 MAC-VRFs/BTs for only locally configured subnets.<a href="#section-4-5" class="pilcrow">¶</a></p>
1830 <p id="section-4-6">
1831 In asymmetric IRB, the lookup operation is asymmetric and the ingress
1832 PE performs three lookups, whereas the egress PE performs a single
1833 lookup -- i.e., the ingress PE performs a MAC lookup, followed by an
1834 IP lookup, followed by a MAC lookup again. The egress PE
1835 performs just a single MAC lookup as depicted in <a href="#fig-3" class="xref">Figure 3</a> below.<a href="#section-4-6" class="pilcrow">¶</a></p>
1836 <span id="name-asymmetric-irb"></span><div id="fig-3">
1837 <figure id="figure-3">
1838 <div class="alignLeft art-text artwork" id="section-4-7.1">
1839 <pre>
1840 Ingress PE Egress PE
1841 +-------------------+ +------------------+
1842 | | | |
1843 | +-&gt; IP-VRF -&gt; | | IP-VRF |
1844 | | | | | |
1845 | BT1 BT2 | | BT3 BT2 |
1846 | | | | | | | |
1847 | | +--|---&gt;----|--------------+ | |
1848 | | | | v |
1849 +-------------------+ +----------------|-+
1850 ^ |
1851 | |
1852 TS1-&gt;-+ +-&gt;-TS2
1853 </pre>
1854 </div>
1855 <figcaption><a href="#figure-3" class="selfRef">Figure 3</a>:
1856 <a href="#name-asymmetric-irb" class="selfRef">Asymmetric IRB</a>
1857 </figcaption></figure>
1858 </div>
1859 <p id="section-4-8">
1860 In asymmetric IRB, as shown in <a href="#fig-3" class="xref">Figure 3</a>, the inter-subnet forwarding between
1861 two PEs is done between their associated MAC-VRFs/BTs.
1862 Therefore, the MPLS or NVO tunnel used for inter-subnet forwarding <span class="bcp14">MUST</span> be
1863 of type Ethernet.
1864
1865 Since only MAC lookup is performed at the egress PE
1866 (e.g., no IP lookup), the TS1's IP packets need to be encapsulated with the
1867 destination TS2's MAC address. In order for the ingress PE to perform such
1868 encapsulation, it needs to maintain TS2's IP and MAC address association in
1869 its ARP table. Furthermore, it needs to maintain destination TS2's MAC
1870 address in the corresponding bridge table even though it may not have any
1871 TSs of the corresponding subnet locally attached. In other words, each PE
1872 participating in asymmetric IRB <span class="bcp14">MUST</span> maintain ARP entries for remote hosts
1873 (hosts connected to other PEs) as well as maintain MAC-VRFs/BTs
1874 and IRB interfaces for ALL subnets in an IP-VRF, including subnets that may
1875 not be locally attached. Therefore, careful consideration of the PE scale
1876 aspects for its ARP table size, its IRB interfaces, and the number and size of its
1877 bridge tables should be given for the application of asymmetric IRB.<a href="#section-4-8" class="pilcrow">¶</a></p>
1878 <p id="section-4-9">
1879 It should be noted that whenever a PE performs a host IP lookup for a
1880 packet that is routed, the IPv4 Time To Live (TTL) or IPv6 hop limit for that packet is
1881 decremented by one, and if it reaches zero, the packet is discarded.
1882 In the case of symmetric IRB, the TTL / hop limit is decremented by
1883 both ingress and egress PEs (once by each), whereas in the case of
1884 asymmetric IRB, the TTL / hop limit is decremented only once by the
1885 ingress PE.<a href="#section-4-9" class="pilcrow">¶</a></p>
1886 <p id="section-4-10">
1887 The following sections define the control and data plane procedures
1888 for symmetric and asymmetric IRB on ingress and egress PEs. The
1889 following figure is used to describe these procedures, showing a
1890 single IP-VRF and a number of BDs on each PE for a
1891 given tenant. That is, an IP-VRF connects one or more EVIs, and each EVI
1892 contains one MAC-VRF; each MAC VRF consists of one or more bridge
1893 tables, one per BD; and a PE has an associated IRB
1894 interface for each BD.<a href="#section-4-10" class="pilcrow">¶</a></p>
1895 <span id="name-irb-forwarding"></span><div id="fig-4">
1896 <figure id="figure-4">
1897 <div class="alignLeft art-text artwork" id="section-4-11.1">
1898 <pre>
1899 PE 1 +---------+
1900 +-------------+ | |
1901 TS1-----| MACx| | | PE2
1902 (M1/IP1) |(BT1) | | | +-------------+
1903 TS5-----| \ | | MPLS/ | |MACy (BT3) |-----TS3
1904 (M5/IP5) |IPx/Mx \ | | VXLAN/ | | / | (M3/IP3)
1905 | (IP-VRF1)|----| NVGRE |---|(IP-VRF1) |
1906 | / | | | | \ |
1907 TS2-----|(BT2) / | | | | (BT1) |-----TS4
1908 (M2/IP2) | | | | | | (M4/IP4)
1909 +-------------+ | | +-------------+
1910 | |
1911 +---------+
1912 </pre>
1913 </div>
1914 <figcaption><a href="#figure-4" class="selfRef">Figure 4</a>:
1915 <a href="#name-irb-forwarding" class="selfRef">IRB Forwarding</a>
1916 </figcaption></figure>
1917 </div>
1918 <div id="sect-4.1">
1919 <section id="section-4.1">
1920 <h3 id="name-irb-interface-and-its-mac-a">
1921 <a href="#section-4.1" class="section-number selfRef">4.1. </a><a href="#name-irb-interface-and-its-mac-a" class="section-name selfRef">IRB Interface and Its MAC and IP Addresses</a>
1922 </h3>
1923 <p id="section-4.1-1">
1924 To support inter-subnet forwarding on a PE, the PE acts as an IP
1925 default gateway from the perspective of the attached Tenant Systems
1926 where default gateway MAC and IP addresses are configured on each IRB
1927 interface associated with its subnet and fall into one of the
1928 following two options:<a href="#section-4.1-1" class="pilcrow">¶</a></p>
1929 <ol start="1" type="1" class="normal type-1" id="section-4.1-2">
1930 <li id="section-4.1-2.1">
1931 <div id="opt1">All the PEs for a given tenant subnet use the same anycast
1932 default gateway IP and MAC addresses. On each PE, these default
1933 gateway IP and MAC addresses correspond to the IRB interface
1934 connecting the bridge table associated with the tenant's VLAN to
1935 the corresponding tenant's IP-VRF.<a href="#opt1" class="pilcrow">¶</a>
1936 </div>
1937 </li>
1938 <li id="section-4.1-2.2">
1939 <div id="opt2">Each PE for a given tenant subnet uses the same anycast default
1940 gateway IP address but its own MAC address. These MAC addresses
1941 are aliased to the same anycast default gateway IP address
1942 through the use of the Default Gateway extended community as
1943 specified in <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span>, which is carried in the EVPN MAC/IP
1944 Advertisement routes. On each PE, this default gateway IP
1945 address, along with its associated MAC addresses, correspond to the
1946 IRB interface connecting the bridge table associated with the
1947 tenant's VLAN to the corresponding tenant's IP-VRF.<a href="#opt2" class="pilcrow">¶</a>
1948 </div>
1949 </li>
1950 </ol>
1951 <p id="section-4.1-3">
1952 It is worth noting that if the applications that are running on the
1953 TSs are employing or relying on any form of MAC security, then the
1954 first option (i.e., using an anycast MAC address) should be used to
1955 ensure that the applications receive traffic from the same IRB
1956 interface MAC address to which they are sending. If the second option
1957 is used, then the IRB interface MAC address <span class="bcp14">MUST</span> be the one used in
1958 the initial ARP reply or ND Neighbor Advertisement (NA) for that TS.<a href="#section-4.1-3" class="pilcrow">¶</a></p>
1959 <p id="section-4.1-4">
1960 Although both of these options are applicable to both symmetric and
1961 asymmetric IRB, <a href="#opt1" class="xref">option 1</a> is recommended because of the ease of
1962 anycast MAC address provisioning on not only the IRB interface
1963 associated with a given subnet across all the PEs corresponding to
1964 that VLAN but also on all IRB interfaces associated with all the
1965 tenant's subnets across all the PEs corresponding to all the VLANs
1966 for that tenant. Furthermore, it simplifies the operation as there
1967 is no need for Default Gateway extended community advertisement and
1968 its associated MAC aliasing procedure. Yet another advantage is that
1969 following host mobility, the host does not need to refresh the
1970 default GW ARP/ND entry.<a href="#section-4.1-4" class="pilcrow">¶</a></p>
1971 <p id="section-4.1-5">
1972 If <a href="#opt1" class="xref">option 1</a> is used, an implementation <span class="bcp14">MAY</span> choose to auto-derive the
1973 anycast MAC address. If auto-derivation is used, the anycast MAC
1974 <span class="bcp14">MUST</span> be auto-derived out of the following ranges (which are defined
1975 in <span>[<a href="#RFC5798" class="xref">RFC5798</a>]</span>):<a href="#section-4.1-5" class="pilcrow">¶</a></p>
1976 <ul class="normal">
1977 <li class="normal" id="section-4.1-6.1">Anycast IPv4 IRB case: 00-00-5E-00-01-{VRID}<a href="#section-4.1-6.1" class="pilcrow">¶</a>
1978 </li>
1979 <li class="normal" id="section-4.1-6.2">Anycast IPv6 IRB case: 00-00-5E-00-02-{VRID}<a href="#section-4.1-6.2" class="pilcrow">¶</a>
1980 </li>
1981 </ul>
1982 <p id="section-4.1-7">
1983 Where the last octet is generated based on a configurable Virtual Router ID
1984 (VRID) (range 1-255). If not explicitly configured, the default value for
1985 the VRID octet is '1'. Auto-derivation of the anycast MAC can only be used
1986 if there is certainty that the auto-derived MAC does not collide with any
1987 customer MAC address.<a href="#section-4.1-7" class="pilcrow">¶</a></p>
1988 <p id="section-4.1-8">
1989 In addition to IP anycast addresses, IRB interfaces can be configured
1990 with non-anycast IP addresses for the purpose of OAM (such as sending a traceroute/ping to these interfaces) for both symmetric and
1991 asymmetric IRB. These IP addresses need to be distributed as VPN
1992 routes when PEs operate in symmetric IRB mode. However, they don't
1993 need to be distributed if the PEs are operating in asymmetric IRB
1994 mode as the non-anycast IP addresses are configured along with their
1995 individual MACs, and they get distributed via the EVPN route type 2
1996 advertisement.<a href="#section-4.1-8" class="pilcrow">¶</a></p>
1997 <p id="section-4.1-9">
1998 For <a href="#opt1" class="xref">option 1</a> -- irrespective of whether only the anycast MAC address or
1999 both anycast and non-anycast MAC addresses (where the latter one is
2000 used for the purpose of OAM) are used on the same IRB -- when a TS sends an ARP
2001 request or ND Neighbor Solicitation (NS) to the PE to which it is attached, the request is sent for the anycast IP address of the IRB
2002 interface associated with the TS's subnet. The reply will use
2003 an anycast MAC address (in both the source MAC in the Ethernet header and
2004 sender hardware address in the payload). For example, in <a href="#fig-4" class="xref">Figure 4</a>,
2005 TS1 is configured with the anycast IPx address as its default gateway
2006 IP address; thus, when it sends an ARP request for IPx (anycast IP
2007 address of the IRB interface for BT1), the PE1 sends an ARP reply
2008 with the MACx, which is the anycast MAC address of that IRB interface.
2009 Traffic routed from IP-VRF1 to TS1 uses the anycast MAC address as the
2010 source MAC address.<a href="#section-4.1-9" class="pilcrow">¶</a></p>
2011 </section>
2012 </div>
2013 <div id="sect-4.2">
2014 <section id="section-4.2">
2015 <h3 id="name-operational-considerations">
2016 <a href="#section-4.2" class="section-number selfRef">4.2. </a><a href="#name-operational-considerations" class="section-name selfRef">Operational Considerations</a>
2017 </h3>
2018 <p id="section-4.2-1">
2019 Symmetric and asymmetric IRB modes may coexist in the same network, and an
2020 ingress PE that supports both forwarding modes for a given tenant can
2021 interwork with egress PEs that support either IRB mode. The egress PE will
2022 indicate the desired forwarding mode for a given host based on the presence
2023 of the Label2 field and the IP-VRF Route Target in the EVPN MAC/IP
2024 Advertisement route. If the Label2 field of the received MAC/IP
2025 Advertisement route for host H1 is non-zero, and one of its Route Targets
2026 identifies the IP-VRF, the ingress PE will use symmetric IRB mode when
2027 forwarding packets destined to H1. If the Label2 field is zero and the
2028 MAC/IP Advertisement route for H1 does not carry any Route Target that
2029 identifies the IP-VRF, the ingress PE will use asymmetric mode when
2030 forwarding traffic to H1.<a href="#section-4.2-1" class="pilcrow">¶</a></p>
2031 <p id="section-4.2-2">
2032 As an example that illustrates the previous statement, suppose PE1
2033 and PE2 need to forward packets from TS2 to TS4 in
2034 <a href="#fig-4" class="xref">Figure 4</a>. Since both PEs are attached to the bridge table of the
2035 destination host, symmetric and asymmetric IRB modes are both
2036 possible as long as the ingress PE, PE1, supports both modes. The
2037 forwarding mode will depend on the mode configured in the egress PE,
2038 PE2. That is:<a href="#section-4.2-2" class="pilcrow">¶</a></p>
2039 <ol start="1" type="1" class="normal type-1" id="section-4.2-3">
2040 <li id="section-4.2-3.1">If PE2 is configured for symmetric IRB mode, PE2 will advertise TS4
2041 MAC/IP addresses in a MAC/IP Advertisement route with a non-zero Label2
2042 field, e.g., Label2 = Lx, and a Route Target that identifies IP-VRF1 in
2043 PE1. IP4 will be installed in PE1's IP-VRF1; TS4's ARP and MAC
2044 information will also be installed in PE1's IRB interface ARP table and
2045 BT1, respectively. When a packet from TS2 destined to TS4 is looked up
2046 in PE1's IP-VRF route table, a longest prefix match lookup will find
2047 IP4 in the IP-VRF, and PE1 will forward using the symmetric IRB mode
2048 and Label Lx.<a href="#section-4.2-3.1" class="pilcrow">¶</a>
2049 </li>
2050 <li id="section-4.2-3.2">However, if PE2 is configured for asymmetric IRB mode, PE2 will
2051 advertise TS4 MAC/IP information in a MAC/IP Advertisement route
2052 with a zero Label2 field and no Route Target identifying IP-VRF1.
2053 In this case, PE2 will install TS4 information in its ARP table
2054 and BT1. When a packet from TS2 to TS4 arrives at PE1, a longest
2055 prefix match on IP-VRF1's route table will yield the local IRB
2056 interface to BT1, where a subsequent ARP and bridge table lookup
2057 will provide the information for an asymmetric forwarding mode to
2058 PE2.<a href="#section-4.2-3.2" class="pilcrow">¶</a>
2059 </li>
2060 </ol>
2061 <p id="section-4.2-4">
2062 Refer to <span>[<a href="#I-D.ietf-bess-evpn-modes-interop" class="xref">EVPN</a>]</span> for more information
2063 about interoperability between symmetric and asymmetric forwarding
2064 modes.<a href="#section-4.2-4" class="pilcrow">¶</a></p>
2065 <p id="section-4.2-5">
2066 The choice between symmetric or asymmetric mode is based on the
2067 operator's preference, and it is a trade-off between scale (which is better in
2068 the symmetric IRB mode) and control plane simplicity (asymmetric IRB
2069 mode simplifies the control plane). In cases where a tenant has
2070 hosts for every subnet attached to all (or most of) the PEs, the ARP and
2071 MAC entries need to be learned by all PEs anyway; therefore, the
2072 asymmetric IRB mode simplifies the forwarding model and saves space
2073 in the IP-VRF route table, since host routes are not installed in the
2074 route table. However, if the tenant does not need to stretch subnets
2075 (broadcast domains) to multiple PEs and inter-subnet forwarding is
2076 needed, the symmetric IRB model will save ARP and bridge table space
2077 in all the PEs (in comparison with the asymmetric IRB model).<a href="#section-4.2-5" class="pilcrow">¶</a></p>
2078 </section>
2079 </div>
2080 </section>
2081 </div>
2082 <div id="sect-5">
2083 <section id="section-5">
2084 <h2 id="name-symmetric-irb-procedures">
2085 <a href="#section-5" class="section-number selfRef">5. </a><a href="#name-symmetric-irb-procedures" class="section-name selfRef">Symmetric IRB Procedures</a>
2086 </h2>
2087 <div id="sect-5.1">
2088 <section id="section-5.1">
2089 <h3 id="name-control-plane-advertising-p">
2090 <a href="#section-5.1" class="section-number selfRef">5.1. </a><a href="#name-control-plane-advertising-p" class="section-name selfRef">Control Plane - Advertising PE</a>
2091 </h3>
2092 <p id="section-5.1-1">
2093 When a PE (e.g., PE1 in <a href="#fig-4" class="xref">Figure 4</a> above) learns the MAC and IP address of
2094 a TS (e.g., via an ARP request or Neighbor Solicitation), it adds the
2095 MAC address to the corresponding MAC-VRF/BT of that
2096 tenant's subnet and adds the IP address to the IP-VRF for that
2097 tenant. Furthermore, it adds this TS's MAC and IP address
2098 association to its ARP table or Neighbor Discovery
2099 Protocol (NDP) cache. It then builds an EVPN
2100 MAC/IP Advertisement route (type 2) as follows and advertises it to
2101 other PEs participating in that tenant's VPN.<a href="#section-5.1-1" class="pilcrow">¶</a></p>
2102 <ul class="normal">
2103 <li class="normal" id="section-5.1-2.1">The Length field of the BGP EVPN Network Layer Reachability Information (NLRI) for an EVPN MAC/IP
2104 Advertisement route <span class="bcp14">MUST</span> be either 40 (if the IPv4 address is carried)
2105 or 52 (if the IPv6 address is carried).<a href="#section-5.1-2.1" class="pilcrow">¶</a>
2106 </li>
2107 <li class="normal" id="section-5.1-2.2">The Route Distinguisher (RD), Ethernet Segment Identifier, Ethernet
2108 Tag ID, MAC Address Length, MAC Address, IP Address Length, IP
2109 Address, and MPLS Label1 fields <span class="bcp14">MUST</span> be set per <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span> and
2110 <span>[<a href="#RFC8365" class="xref">RFC8365</a>]</span>.<a href="#section-5.1-2.2" class="pilcrow">¶</a>
2111 </li>
2112 <li class="normal" id="section-5.1-2.3">The MPLS Label2 field is set to either an MPLS label or a VNI
2113 corresponding to the tenant's IP-VRF. In the case of an MPLS
2114 label, this field is encoded as 3 octets, where the high-order 20
2115 bits contain the label value.<a href="#section-5.1-2.3" class="pilcrow">¶</a>
2116 </li>
2117 </ul>
2118 <p id="section-5.1-3">
2119 Just as in <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span>, the RD, Ethernet Tag ID, MAC Address Length,
2120 MAC Address, IP Address Length, and IP Address fields are part of the
2121 route key used by BGP to compare routes. The rest of the fields are
2122 not part of the route key.<a href="#section-5.1-3" class="pilcrow">¶</a></p>
2123 <p id="section-5.1-4">
2124 This route is advertised along with the following two extended
2125 communities:<a href="#section-5.1-4" class="pilcrow">¶</a></p>
2126 <ol start="1" type="1" class="normal type-1" id="section-5.1-5">
2127 <li id="section-5.1-5.1">Encapsulation Extended Community<a href="#section-5.1-5.1" class="pilcrow">¶</a>
2128 </li>
2129 <li id="section-5.1-5.2">EVPN Router's MAC Extended Community<a href="#section-5.1-5.2" class="pilcrow">¶</a>
2130 </li>
2131 </ol>
2132 <p id="section-5.1-6">
2133 This route is advertised with one or more Encapsulation Extended
2134 Communities <span>[<a href="#RFC9012" class="xref">RFC9012</a>]</span>, one for each encapsulation type supported by
2135 the advertising PE. If one or more encapsulation types require an
2136 Ethernet frame, a single EVPN Router's MAC Extended Community (<a href="#sect-8.1" class="xref">Section 8.1</a>) is also advertised. This extended community specifies the MAC
2137 address to be used as the inner destination MAC address in an
2138 Ethernet frame sent to the advertising PE.<a href="#section-5.1-6" class="pilcrow">¶</a></p>
2139 <p id="section-5.1-7">
2140 This route <span class="bcp14">MUST</span> be advertised with two Route Targets, one
2141 corresponding to the MAC-VRF of the tenant's subnet and another
2142 corresponding to the tenant's IP-VRF.<a href="#section-5.1-7" class="pilcrow">¶</a></p>
2143 </section>
2144 </div>
2145 <div id="sect-5.2">
2146 <section id="section-5.2">
2147 <h3 id="name-control-plane-receiving-pe">
2148 <a href="#section-5.2" class="section-number selfRef">5.2. </a><a href="#name-control-plane-receiving-pe" class="section-name selfRef">Control Plane - Receiving PE</a>
2149 </h3>
2150 <p id="section-5.2-1">
2151 When a PE (e.g., PE2 in <a href="#fig-4" class="xref">Figure 4</a> above) receives this EVPN MAC/IP
2152 Advertisement route, it performs the following:<a href="#section-5.2-1" class="pilcrow">¶</a></p>
2153 <ul class="normal">
2154 <li class="normal" id="section-5.2-2.1">The MAC-VRF Route Target and Ethernet Tag,
2155 if the latter is non-zero, are used to identify the correct MAC-VRF
2156 and bridge table, and if they are found, the MAC address is imported.
2157 The IP-VRF Route Target is used to identify the correct IP-VRF, and if
2158 it is found, the IP address is imported.<a href="#section-5.2-2.1" class="pilcrow">¶</a>
2159 </li>
2160 </ul>
2161 <p id="section-5.2-3">
2162 If the MPLS Label2 field is non-zero, it means that this route is to
2163 be used for symmetric IRB, and the MPLS label2 value is to be used
2164 when sending a packet for this IP address to the advertising PE.<a href="#section-5.2-3" class="pilcrow">¶</a></p>
2165 <p id="section-5.2-4">
2166 If the receiving PE supports asymmetric IRB mode and receives this route with both the MAC-VRF and IP-VRF Route Targets but the MAC/IP Advertisement route does not include the MPLS
2167 Label2 field, then the receiving PE installs the MAC address in the corresponding MAC-VRF and the (IP,
2168 MAC) association in the ARP table for that tenant (identified by the
2169 corresponding IP-VRF Route Target).<a href="#section-5.2-4" class="pilcrow">¶</a></p>
2170 <p id="section-5.2-5">
2171 If the receiving PE receives this route with both the MAC-VRF and IP-VRF
2172 Route Targets, and if the receiving PE does not support either asymmetric or
2173 symmetric IRB modes but has the corresponding MAC-VRF, then it only
2174 imports the MAC address.<a href="#section-5.2-5" class="pilcrow">¶</a></p>
2175 <p id="section-5.2-6">
2176 If the receiving PE receives this route with both the MAC-VRF and IP-VRF
2177 Route Targets and the MAC/IP Advertisement route includes the MPLS Label2 field
2178 but the receiving PE only supports asymmetric IRB mode, then the receiving
2179 PE <span class="bcp14">MUST</span> ignore the MPLS Label2 field and install the MAC address in the
2180 corresponding MAC-VRF and (IP, MAC) association in the ARP table for that
2181 tenant (identified by the corresponding IP-VRF Route Target).<a href="#section-5.2-6" class="pilcrow">¶</a></p>
2182 </section>
2183 </div>
2184 <div id="sect-5.3">
2185 <section id="section-5.3">
2186 <h3 id="name-subnet-route-advertisement">
2187 <a href="#section-5.3" class="section-number selfRef">5.3. </a><a href="#name-subnet-route-advertisement" class="section-name selfRef">Subnet Route Advertisement</a>
2188 </h3>
2189 <p id="section-5.3-1">
2190 In the case of symmetric IRB, a Layer 3 subnet and IRB interface
2191 corresponding to a MAC-VRF/BT are required to be provisioned at a
2192 PE only if that PE has locally attached hosts in that subnet. In order to
2193 enable inter-subnet routing across PEs in a deployment where not all
2194 subnets are provisioned at all PEs participating in an EVPN IRB instance,
2195 PEs <span class="bcp14">MUST</span> advertise local subnet routes as EVPN RT-5. These subnet routes
2196 are required for bootstrapping host (IP, MAC) learning using gleaning
2197 procedures initiated by an inter-subnet data packet.<a href="#section-5.3-1" class="pilcrow">¶</a></p>
2198 <p id="section-5.3-2">
2199 That is, if a given host's (IP, MAC) association is unknown, and an
2200 ingress PE needs to send a packet to that host, then that ingress PE
2201 needs to know which egress PEs are attached to the subnet in which
2202 the host resides in order to send the packet to one of those PEs,
2203 causing the PE receiving the packet to probe for that host. For
2204 example, consider a subnet A that is locally attached to PE1 and
2205 subnet B that is locally attached to PE2 and PE3. Host A in
2206 subnet A, which is attached to PE1, initiates a data packet destined to
2207 host B in subnet B, which is attached to PE3. If host B's (IP, MAC)
2208 has not yet been learned via either a gratuitous ARP OR a prior
2209 gleaning procedure, a new gleaning procedure <span class="bcp14">MUST</span> be triggered for
2210 host B's (IP, MAC) to be learned and advertised across the EVPN
2211 network. Since host B's subnet is not local to PE1, an IP lookup for
2212 host B at PE1 will not trigger this gleaning procedure for host B's
2213 (IP, MAC). Therefore, PE1 <span class="bcp14">MUST</span> learn subnet B's prefix route via
2214 EVPN RT-5 advertised from PE2 and PE3, so it can route the packet to
2215 one of the PEs that have subnet B locally attached. Once the packet
2216 is received at PE2 OR PE3, and the route lookup yields a glean
2217 result, an ARP request is triggered and flooded across the Layer 2
2218 overlay.
2219
2220 This ARP request would be received and replied to by host
2221 B, resulting in host B (IP, MAC) learning at PE3 and its
2222 advertisement across the EVPN network. Packets from host A to host B
2223 can now be routed directly from PE1 to PE3. Advertisement of local
2224 subnet EVPN RT-5 for an IP-VRF <span class="bcp14">MAY</span> typically be achieved via
2225 provisioning-connected route redistribution to BGP.<a href="#section-5.3-2" class="pilcrow">¶</a></p>
2226 </section>
2227 </div>
2228 <div id="sect-5.4">
2229 <section id="section-5.4">
2230 <h3 id="name-data-plane-ingress-pe">
2231 <a href="#section-5.4" class="section-number selfRef">5.4. </a><a href="#name-data-plane-ingress-pe" class="section-name selfRef">Data Plane - Ingress PE</a>
2232 </h3>
2233 <p id="section-5.4-1">
2234 When an Ethernet frame is received by an ingress PE (e.g., PE1 in
2235 <a href="#fig-4" class="xref">Figure 4</a> above), the PE uses the AC ID (e.g., VLAN ID) to identify
2236 the associated MAC-VRF/BT, and it performs a lookup on the
2237 destination MAC address. If the MAC address corresponds to its IRB
2238 interface MAC address, the ingress PE deduces that the packet must be
2239 inter-subnet routed. Hence, the ingress PE performs an IP lookup in
2240 the associated IP-VRF table. The lookup identifies the BGP next hop of the egress PE along with the tunnel/encapsulation type and the associated
2241 MPLS/VNI values. The ingress PE also decrements the TTL / hop limit
2242 for that packet by one, and if it reaches zero, the ingress PE
2243 discards the packet.<a href="#section-5.4-1" class="pilcrow">¶</a></p>
2244 <p id="section-5.4-2">
2245 If the tunnel type is that of an MPLS or IP-only NVO tunnel, then the TS's
2246 IP packet is sent over the tunnel without any Ethernet header.
2247 However, if the tunnel type is that of an Ethernet NVO tunnel, then an
2248 Ethernet header needs to be added to the TS's IP packet. The source
2249 MAC address of this inner Ethernet header is set to the ingress PE's
2250 router MAC address, and the destination MAC address of this inner
2251 Ethernet header is set to the egress PE's router MAC address learned
2252 via the EVPN Router's MAC Extended Community attached to the route. The MPLS VPN
2253 label is set to the received label2 in the route. In the case of the Ethernet NVO tunnel type, the VNI may be set one of two ways:<a href="#section-5.4-2" class="pilcrow">¶</a></p>
2254 <span class="break"></span><dl class="dlParallel" id="section-5.4-3">
2255 <dt id="section-5.4-3.1">downstream mode:</dt>
2256 <dd style="margin-left: 1.5em" id="section-5.4-3.2">The VNI is set to the received label2 in the route,
2257 which is downstream assigned.<a href="#section-5.4-3.2" class="pilcrow">¶</a>
2258 </dd>
2259 <dd class="break"></dd>
2260 <dt id="section-5.4-3.3">global mode:</dt>
2261 <dd style="margin-left: 1.5em" id="section-5.4-3.4">The VNI is set to the received label2 in the route, which
2262 is assigned domain-wide. This VNI value from the received label2 <span class="bcp14">MUST</span>
2263 be the same as the locally configured VNI for the IP-VRF as all
2264 PEs in the NVO <span class="bcp14">MUST</span> be configured with the same IP-VRF VNI for
2265 this mode of operation. If the received label2 value does not
2266 match the locally configured VNI value, the route <span class="bcp14">MUST NOT</span> be used,
2267 and an error message <span class="bcp14">SHOULD</span> be logged.<a href="#section-5.4-3.4" class="pilcrow">¶</a>
2268 </dd>
2269 <dd class="break"></dd>
2270 </dl>
2271 <p id="section-5.4-4">
2272 PEs may be configured to operate in one of these two modes depending
2273 on the administrative domain boundaries across PEs participating in
2274 the NVO and the PE's capability to support downstream VNI mode.<a href="#section-5.4-4" class="pilcrow">¶</a></p>
2275 <p id="section-5.4-5">
2276 In the case of NVO tunnel encapsulation, the outer source and
2277 destination IP addresses are set to the ingress and egress PE BGP
2278 next-hop IP addresses, respectively.<a href="#section-5.4-5" class="pilcrow">¶</a></p>
2279 </section>
2280 </div>
2281 <div id="sect-5.5">
2282 <section id="section-5.5">
2283 <h3 id="name-data-plane-egress-pe">
2284 <a href="#section-5.5" class="section-number selfRef">5.5. </a><a href="#name-data-plane-egress-pe" class="section-name selfRef">Data Plane - Egress PE</a>
2285 </h3>
2286 <p id="section-5.5-1">
2287 When the tenant's MPLS or NVO encapsulated packet is received over an
2288 MPLS or NVO tunnel by the egress PE, the egress PE removes the NVO tunnel
2289 encapsulation and uses the VPN MPLS label (for MPLS encapsulation) or
2290 VNI (for NVO encapsulation) to identify the IP-VRF in which IP lookup
2291 needs to be performed. If the VPN MPLS label or VNI identifies a
2292 MAC-VRF instead of an IP-VRF, then the procedures in <a href="#sect-6.4" class="xref">Section 6.4</a> for
2293 asymmetric IRB are executed.<a href="#section-5.5-1" class="pilcrow">¶</a></p>
2294 <p id="section-5.5-2">
2295 The lookup in the IP-VRF identifies a local adjacency to the IRB
2296 interface associated with the egress subnet's MAC-VRF/BT.
2297 The egress PE also decrements the TTL / hop limit for that packet by
2298 one, and if it reaches zero, the egress PE discards the packet.<a href="#section-5.5-2" class="pilcrow">¶</a></p>
2299 <p id="section-5.5-3">
2300 The egress PE gets the destination TS's MAC address for that TS's IP
2301 address from its ARP table or NDP cache. It encapsulates the packet
2302 with that destination MAC address and a source MAC address
2303 corresponding to that IRB interface and sends the packet to its
2304 destination subnet MAC-VRF/BT.<a href="#section-5.5-3" class="pilcrow">¶</a></p>
2305 <p id="section-5.5-4">
2306 The destination MAC address lookup in the MAC-VRF/BT
2307 results in the local adjacency (e.g., local interface) over which the
2308 Ethernet frame is sent.<a href="#section-5.5-4" class="pilcrow">¶</a></p>
2309 </section>
2310 </div>
2311 </section>
2312 </div>
2313 <div id="sect-6">
2314 <section id="section-6">
2315 <h2 id="name-asymmetric-irb-procedures">
2316 <a href="#section-6" class="section-number selfRef">6. </a><a href="#name-asymmetric-irb-procedures" class="section-name selfRef">Asymmetric IRB Procedures</a>
2317 </h2>
2318 <div id="sect-6.1">
2319 <section id="section-6.1">
2320 <h3 id="name-control-plane-advertising-pe">
2321 <a href="#section-6.1" class="section-number selfRef">6.1. </a><a href="#name-control-plane-advertising-pe" class="section-name selfRef">Control Plane - Advertising PE</a>
2322 </h3>
2323 <p id="section-6.1-1">
2324 When a PE (e.g., PE1 in <a href="#fig-4" class="xref">Figure 4</a> above) learns the MAC and IP address of
2325 an attached TS (e.g., via an ARP request or ND Neighbor
2326 Solicitation), it populates its MAC-VRF/BT, IP-VRF, and ARP
2327 table or NDP cache just as in the case for symmetric IRB. It then
2328 builds an EVPN MAC/IP Advertisement route (type 2) as follows and
2329 advertises it to other PEs participating in that tenant's VPN.<a href="#section-6.1-1" class="pilcrow">¶</a></p>
2330 <ul class="normal">
2331 <li class="normal" id="section-6.1-2.1">The Length field of the BGP EVPN NLRI for an EVPN MAC/IP
2332 Advertisement route <span class="bcp14">MUST</span> be either 37 (if an IPv4 address is carried)
2333 or 49 (if an IPv6 address is carried).<a href="#section-6.1-2.1" class="pilcrow">¶</a>
2334 </li>
2335 <li class="normal" id="section-6.1-2.2">The RD, Ethernet Segment Identifier, Ethernet
2336 Tag ID, MAC Address Length, MAC Address, IP Address Length, IP
2337 Address, and MPLS Label1 fields <span class="bcp14">MUST</span> be set per <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span> and
2338 <span>[<a href="#RFC8365" class="xref">RFC8365</a>]</span>.<a href="#section-6.1-2.2" class="pilcrow">¶</a>
2339 </li>
2340 <li class="normal" id="section-6.1-2.3">The MPLS Label2 field <span class="bcp14">MUST NOT</span> be included in this route.<a href="#section-6.1-2.3" class="pilcrow">¶</a>
2341 </li>
2342 </ul>
2343 <p id="section-6.1-3">
2344 Just as in <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span>, the RD, Ethernet Tag ID, MAC Address Length,
2345 MAC Address, IP Address Length, and IP Address fields are part of the
2346 route key used by BGP to compare routes. The rest of the fields are
2347 not part of the route key.<a href="#section-6.1-3" class="pilcrow">¶</a></p>
2348 <p id="section-6.1-4">
2349 This route is advertised along with the following extended community:<a href="#section-6.1-4" class="pilcrow">¶</a></p>
2350 <ul class="normal">
2351 <li class="normal" id="section-6.1-5.1">Tunnel Type Extended Community<a href="#section-6.1-5.1" class="pilcrow">¶</a>
2352 </li>
2353 </ul>
2354 <p id="section-6.1-6">
2355 For asymmetric IRB mode, the EVPN Router's MAC Extended Community is not
2356 needed because forwarding is performed using destination TS's MAC
2357 address, which is carried in this EVPN route type 2 advertisement.<a href="#section-6.1-6" class="pilcrow">¶</a></p>
2358 <p id="section-6.1-7">
2359 This route <span class="bcp14">MUST</span> always be advertised with the MAC-VRF Route Target.
2360 It <span class="bcp14">MAY</span> also be advertised with a second Route Target corresponding to
2361 the IP-VRF.<a href="#section-6.1-7" class="pilcrow">¶</a></p>
2362 </section>
2363 </div>
2364 <div id="sect-6.2">
2365 <section id="section-6.2">
2366 <h3 id="name-control-plane-receiving-pe-2">
2367 <a href="#section-6.2" class="section-number selfRef">6.2. </a><a href="#name-control-plane-receiving-pe-2" class="section-name selfRef">Control Plane - Receiving PE</a>
2368 </h3>
2369 <p id="section-6.2-1">
2370 When a PE (e.g., PE2 in <a href="#fig-4" class="xref">Figure 4</a> above) receives this EVPN MAC/IP
2371 Advertisement route, it performs the following:<a href="#section-6.2-1" class="pilcrow">¶</a></p>
2372 <ul class="normal">
2373 <li class="normal" id="section-6.2-2.1">Using the MAC-VRF Route Target, it identifies
2374 the corresponding MAC-VRF and imports the MAC address into it. For
2375 asymmetric IRB mode, it is assumed that all PEs participating in a
2376 tenant's VPN are configured with all subnets (i.e., all VLANs) and
2377 corresponding MAC-VRFs/BTs even if there are no locally
2378 attached TSs for some of these subnets. This is because the ingress PE needs to do forwarding based on the destination TS's MAC address
2379 and perform NVO tunnel encapsulation as the property of a lookup in the MAC-VRF/BT.<a href="#section-6.2-2.1" class="pilcrow">¶</a>
2380 </li>
2381 <li class="normal" id="section-6.2-2.2">If only the MAC-VRF Route Target is used, then the receiving PE uses
2382 the MAC-VRF Route Target to identify the corresponding IP-VRF --
2383 i.e., many MAC-VRF Route Targets map to the same IP-VRF for a
2384 given tenant. In this case, MAC-VRF may be used by the receiving
2385 PE to identify the corresponding IP-VRF via the IRB interface
2386 associated with the subnet MAC-VRF/BT. In this case,
2387 the MAC-VRF Route Target may be used by the receiving PE to
2388 identify the corresponding IP-VRF.<a href="#section-6.2-2.2" class="pilcrow">¶</a>
2389 </li>
2390 <li class="normal" id="section-6.2-2.3">Using the MAC-VRF Route Target, the receiving PE identifies the
2391 corresponding ARP table or NDP cache for the tenant, and it adds an
2392 entry to the ARP table or NDP cache for the TS's MAC and IP
2393 address association. It should be noted that the tenant's ARP
2394 table or NDP cache at the receiving PE is identified by all the
2395 MAC-VRF Route Targets for that tenant.<a href="#section-6.2-2.3" class="pilcrow">¶</a>
2396 </li>
2397 <li class="normal" id="section-6.2-2.4">If the IP-VRF Route Target is included, it may be used to import the
2398 route to IP-VRF. If the IP-VRF Route Target is not included, MAC-VRF
2399 is used to derive the corresponding IP-VRF for import, as explained in
2400 the prior section. In both cases, an IP-VRF route is installed with
2401 the TS MAC binding included in the received route.<a href="#section-6.2-2.4" class="pilcrow">¶</a>
2402 </li>
2403 </ul>
2404 <p id="section-6.2-3">
2405 If the receiving PE receives the MAC/IP Advertisement route with the MPLS
2406 Label2 field but the receiving PE only supports asymmetric IRB mode,
2407 then the receiving PE <span class="bcp14">MUST</span> ignore the MPLS Label2 field and install the
2408 MAC address in the corresponding MAC-VRF and (IP, MAC) association in
2409 the ARP table or NDP cache for that tenant (with the IRB interface
2410 identified by the MAC-VRF).<a href="#section-6.2-3" class="pilcrow">¶</a></p>
2411 </section>
2412 </div>
2413 <div id="sect-6.3">
2414 <section id="section-6.3">
2415 <h3 id="name-data-plane-ingress-pe-2">
2416 <a href="#section-6.3" class="section-number selfRef">6.3. </a><a href="#name-data-plane-ingress-pe-2" class="section-name selfRef">Data Plane - Ingress PE</a>
2417 </h3>
2418 <p id="section-6.3-1">
2419 When an Ethernet frame is received by an ingress PE (e.g., PE1 in
2420 <a href="#fig-4" class="xref">Figure 4</a> above), the PE uses the AC ID (e.g., VLAN ID) to identify
2421 the associated MAC-VRF/BT, and it performs a lookup on the
2422 destination MAC address. If the MAC address corresponds to its IRB
2423 interface MAC address, the ingress PE deduces that the packet must be
2424 inter-subnet routed. Hence, the ingress PE performs an IP lookup in
2425 the associated IP-VRF table. The lookup identifies a local adjacency
2426 to the IRB interface associated with the egress subnet's MAC-VRF/
2427 bridge table. The ingress PE also decrements the TTL / hop limit for
2428 that packet by one, and if it reaches zero, the ingress PE discards
2429 the packet.<a href="#section-6.3-1" class="pilcrow">¶</a></p>
2430 <p id="section-6.3-2">
2431 The ingress PE gets the destination TS's MAC address for that TS's IP
2432 address from its ARP table or NDP cache. It encapsulates the packet
2433 with that destination MAC address and a source MAC address
2434 corresponding to that IRB interface and sends the packet to its
2435 destination subnet MAC-VRF/BT.<a href="#section-6.3-2" class="pilcrow">¶</a></p>
2436 <p id="section-6.3-3">
2437 The destination MAC address lookup in the MAC-VRF/BT
2438 results in a BGP next-hop address of the egress PE along with label1 (L2
2439 VPN MPLS label or VNI). The ingress PE encapsulates the packet using
2440 the Ethernet NVO tunnel of the choice (e.g., VXLAN or NVGRE) and sends
2441 the packet to the egress PE. Because the packet forwarding is
2442 between the ingress PE's MAC-VRF/BT and the egress PE's MAC-VRF/
2443 bridge table, the packet encapsulation procedures follow that of
2444 <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span> for MPLS and <span>[<a href="#RFC8365" class="xref">RFC8365</a>]</span> for VXLAN encapsulations.<a href="#section-6.3-3" class="pilcrow">¶</a></p>
2445 </section>
2446 </div>
2447 <div id="sect-6.4">
2448 <section id="section-6.4">
2449 <h3 id="name-data-plane-egress-pe-2">
2450 <a href="#section-6.4" class="section-number selfRef">6.4. </a><a href="#name-data-plane-egress-pe-2" class="section-name selfRef">Data Plane - Egress PE</a>
2451 </h3>
2452 <p id="section-6.4-1">
2453 When a tenant's Ethernet frame is received over an NVO tunnel by the
2454 egress PE, the egress PE removes the NVO tunnel encapsulation and uses
2455 the VPN MPLS label (for MPLS encapsulation) or VNI (for NVO
2456 encapsulation) to identify the MAC-VRF/BT in which the MAC
2457 lookup needs to be performed.<a href="#section-6.4-1" class="pilcrow">¶</a></p>
2458 <p id="section-6.4-2">
2459 The MAC lookup results in a local adjacency (e.g., local interface)
2460 over which the packet needs to get sent.<a href="#section-6.4-2" class="pilcrow">¶</a></p>
2461 <p id="section-6.4-3">
2462 Note that the forwarding behavior on the egress PE is the same as the EVPN intra-subnet forwarding described in <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span> for MPLS and
2463 <span>[<a href="#RFC8365" class="xref">RFC8365</a>]</span> for NVO networks. In other words, all the packet
2464 processing associated with the inter-subnet forwarding semantics is
2465 confined to the ingress PE for asymmetric IRB mode.<a href="#section-6.4-3" class="pilcrow">¶</a></p>
2466 <p id="section-6.4-4">
2467 It should also be noted that <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span> provides a different level of
2468 granularity for the EVPN label. Besides identifying the bridge
2469 domain table, it can be used to identify the egress interface or a
2470 destination MAC address on that interface. If an EVPN label is used for
2471 an egress interface or individual MAC address identification, then no
2472 MAC lookup is needed in the egress PE for MPLS encapsulation, and the
2473 packet can be directly forwarded to the egress interface just based
2474 on the EVPN label lookup.<a href="#section-6.4-4" class="pilcrow">¶</a></p>
2475 </section>
2476 </div>
2477 </section>
2478 </div>
2479 <div id="sect-7">
2480 <section id="section-7">
2481 <h2 id="name-mobility-procedure">
2482 <a href="#section-7" class="section-number selfRef">7. </a><a href="#name-mobility-procedure" class="section-name selfRef">Mobility Procedure</a>
2483 </h2>
2484 <p id="section-7-1">
2485 When a TS moves from one NVE (aka source NVE) to another NVE (aka
2486 target NVE), it is important that the MAC Mobility procedures be
2487 properly executed and the corresponding MAC-VRF and IP-VRF tables on
2488 all participating NVEs be updated. <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span> describes the MAC
2489 Mobility procedures for L2-only services for both single-homed TS and
2490 multihomed TS. This section describes the incremental procedures
2491 and BGP Extended Communities needed to handle the MAC Mobility for
2492 IRB. In order to place the emphasis on the differences between
2493 L2-only and IRB use cases, the incremental procedure is described for
2494 a single-homed TS with the expectation that the additional steps needed
2495 for a multihomed TS can be extended per <span><a href="https://www.rfc-editor.org/rfc/rfc7432#section-15" class="relref">Section 15</a> of [<a href="#RFC7432" class="xref">RFC7432</a>]</span>.
2496 This section describes mobility procedures for both symmetric and
2497 asymmetric IRB. Although the language used in this section is for
2498 IPv4 ARP, it equally applies to IPv6 ND.<a href="#section-7-1" class="pilcrow">¶</a></p>
2499 <p id="section-7-2">
2500 When a TS moves from a source NVE to a target NVE, it can behave in
2501 one of the following three ways:<a href="#section-7-2" class="pilcrow">¶</a></p>
2502 <ol start="1" type="1" class="normal type-1" id="section-7-3">
2503 <li id="section-7-3.1">
2504 <div id="way1">TS initiates an ARP request upon a move to the target NVE.<a href="#way1" class="pilcrow">¶</a>
2505 </div>
2506 </li>
2507 <li id="section-7-3.2">
2508 <div id="way2">TS sends a data packet without first initiating an ARP request to
2509 the target NVE.<a href="#way2" class="pilcrow">¶</a>
2510 </div>
2511 </li>
2512 <li id="section-7-3.3">
2513 <div id="way3">TS is a silent host and neither initiates an ARP request nor
2514 sends any packets.<a href="#way3" class="pilcrow">¶</a>
2515 </div>
2516 </li>
2517 </ol>
2518 <p id="section-7-4">
2519 Depending on the expected TS's behavior, an NVE needs to handle at least
2520 the <a href="#way1" class="xref">first</a> option and should be able to handle the <a href="#way2" class="xref">second</a> and <a href="#way3" class="xref">third</a> options.
2521 The following subsections describe the procedures for each scenario where it
2522 is assumed that the MAC and IP addresses of a TS have a one-to-one
2523 relationship (i.e., there is one IP address per MAC address and vice
2524 versa). The procedures for host mobility detection in the presence of
2525 a many-to-one relationship is outside the scope of this document, and it is
2526 covered in <span>[<a href="#I-D.ietf-bess-evpn-irb-extended-mobility" class="xref">EXTENDED-MOBILITY</a>]</span>. The
2527 "many-to-one relationship" refers to many host IP addresses corresponding to a
2528 single host MAC address or many host MAC addresses corresponding to a
2529 single IP address. It should be noted that in the case of IPv6, a link-local
2530 IP address does not count in a many-to-one relationship because that address
2531 is confined to a single Ethernet segment, and it is not used for host mobility
2532 (i.e., by definition, host mobility is between two different Ethernet
2533 segments). Therefore, when an IPv6 host is configured with both a Global
2534 Unicast address (or a Unique Local address) and a link-local address, for
2535 the purpose of host mobility, it is considered with a single IP
2536 address.<a href="#section-7-4" class="pilcrow">¶</a></p>
2537 <div id="sect-7.1">
2538 <section id="section-7.1">
2539 <h3 id="name-initiating-a-gratuitous-arp">
2540 <a href="#section-7.1" class="section-number selfRef">7.1. </a><a href="#name-initiating-a-gratuitous-arp" class="section-name selfRef">Initiating a Gratuitous ARP upon a Move</a>
2541 </h3>
2542 <p id="section-7.1-1">
2543 In this scenario, when a TS moves from a source NVE to a target NVE,
2544 the TS initiates a gratuitous ARP upon the move to the target NVE.<a href="#section-7.1-1" class="pilcrow">¶</a></p>
2545 <p id="section-7.1-2">
2546 The target NVE, upon receiving this ARP message, updates its MAC-VRF,
2547 IP-VRF, and ARP table with the host MAC, IP, and local adjacency
2548 information (e.g., local interface).<a href="#section-7.1-2" class="pilcrow">¶</a></p>
2549 <p id="section-7.1-3">
2550 Since this NVE has previously learned the same MAC and IP addresses
2551 from the source NVE, it recognizes that there has been a MAC move, and
2552 it initiates MAC Mobility procedures per <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span> by advertising an
2553 EVPN MAC/IP Advertisement route with both the MAC and IP addresses
2554 filled in (per Sections <a href="#sect-5.1" class="xref">5.1</a> and <a href="#sect-6.1" class="xref">6.1</a>) along with the MAC Mobility extended
2555 community, with the sequence number incremented by one. The target
2556 NVE also exercises the MAC duplication detection procedure in <span><a href="https://www.rfc-editor.org/rfc/rfc7432#section-15.1" class="relref">Section 15.1</a> of [<a href="#RFC7432" class="xref">RFC7432</a>]</span>.<a href="#section-7.1-3" class="pilcrow">¶</a></p>
2557 <p id="section-7.1-4">
2558 The source NVE, upon receiving this MAC/IP Advertisement route,
2559 realizes that the MAC has moved to the target NVE. It updates its
2560 MAC-VRF and IP-VRF table accordingly with the adjacency information
2561 of the target NVE. In the case of the asymmetric IRB, the source NVE
2562 also updates its ARP table with the received adjacency information,
2563 and in the case of the symmetric IRB, the source NVE removes the
2564 entry associated with the received (IP, MAC) from its local ARP
2565 table. It then withdraws its EVPN MAC/IP Advertisement route.
2566 Furthermore, it sends an ARP probe locally to ensure that the MAC is
2567 gone. If an ARP response is received, the source NVE updates its ARP
2568 entry for that (IP, MAC) and re-advertises an EVPN MAC/IP
2569 Advertisement route for that (IP, MAC) along with the MAC Mobility
2570 extended community, with the sequence number incremented by one. The
2571 source NVE also exercises the MAC duplication detection procedure in
2572 <span><a href="https://www.rfc-editor.org/rfc/rfc7432#section-15.1" class="relref">Section 15.1</a> of [<a href="#RFC7432" class="xref">RFC7432</a>]</span>.<a href="#section-7.1-4" class="pilcrow">¶</a></p>
2573 <p id="section-7.1-5">
2574 All other remote NVE devices, upon receiving the MAC/IP Advertisement route
2575 with the MAC Mobility extended community, compare the sequence number in this
2576 advertisement with the one previously received. If the new sequence number
2577 is greater than the old one, then they update the MAC/IP addresses of the
2578 TS in their corresponding MAC-VRF and IP-VRF tables to point to the target
2579 NVE. Furthermore, upon receiving the MAC/IP withdraw for the TS from the
2580 source NVE, these remote PEs perform the cleanups for their BGP tables.<a href="#section-7.1-5" class="pilcrow">¶</a></p>
2581 </section>
2582 </div>
2583 <div id="sect-7.2">
2584 <section id="section-7.2">
2585 <h3 id="name-sending-data-traffic-withou">
2586 <a href="#section-7.2" class="section-number selfRef">7.2. </a><a href="#name-sending-data-traffic-withou" class="section-name selfRef">Sending Data Traffic without an ARP Request</a>
2587 </h3>
2588 <p id="section-7.2-1">
2589 In this scenario, when a TS moves from a source NVE to a target NVE,
2590 the TS starts sending data traffic without first initiating an ARP
2591 request.<a href="#section-7.2-1" class="pilcrow">¶</a></p>
2592 <p id="section-7.2-2">
2593 The target NVE, upon receiving the first data packet, learns the MAC
2594 address of the TS in the data plane and updates its MAC-VRF table
2595 with the MAC address and the local adjacency information (e.g., local
2596 interface) accordingly. The target NVE realizes that there has been
2597 a MAC move because the same MAC address has been learned remotely
2598 from the source NVE.<a href="#section-7.2-2" class="pilcrow">¶</a></p>
2599 <p id="section-7.2-3">
2600 If EVPN-IRB NVEs are configured to advertise MAC-only routes in
2601 addition to MAC-and-IP EVPN routes, then the following steps are
2602 taken:<a href="#section-7.2-3" class="pilcrow">¶</a></p>
2603 <ul class="normal">
2604 <li class="normal" id="section-7.2-4.1">The target NVE, upon learning this MAC address in the data plane,
2605 updates this MAC address entry in the corresponding MAC-VRF with
2606 the local adjacency information (e.g., local interface). It also
2607 recognizes that this MAC has moved and initiates MAC Mobility
2608 procedures per <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span> by advertising an EVPN MAC/IP
2609 Advertisement route with only the MAC address filled in along with the
2610 MAC Mobility extended community, with the sequence number
2611 incremented by one.<a href="#section-7.2-4.1" class="pilcrow">¶</a>
2612 </li>
2613 <li class="normal" id="section-7.2-4.2">The source NVE, upon receiving this MAC/IP Advertisement route,
2614 realizes that the MAC has moved to the new NVE. It updates its
2615 MAC-VRF table with the adjacency information for that MAC address
2616 to point to the target NVE and withdraws its EVPN MAC/IP
2617 Advertisement route that has only the MAC address (if it has
2618 advertised such a route previously). Furthermore, it searches for
2619 the corresponding MAC-IP entry and sends an ARP probe for this
2620 (IP, MAC) pair. The ARP request message is sent both locally to
2621 all attached TSs in that subnet as well as to other
2622 NVEs participating in that subnet, including the target NVE. Note
2623 that the PE needs to maintain a correlation between MAC and MAC-IP
2624 route entries in the MAC-VRF to accomplish this.<a href="#section-7.2-4.2" class="pilcrow">¶</a>
2625 </li>
2626 <li class="normal" id="section-7.2-4.3">The target NVE passes the ARP request to its locally attached TSs,
2627 and when it receives the ARP response, it updates its IP-VRF and
2628 ARP table with the host (IP, MAC) information. It also sends an
2629 EVPN MAC/IP Advertisement route with both the MAC and IP addresses
2630 filled in along with the MAC Mobility extended community, with the
2631 sequence number set to the same value as the one for the MAC-only
2632 Advertisement route it sent previously.<a href="#section-7.2-4.3" class="pilcrow">¶</a>
2633 </li>
2634 <li class="normal" id="section-7.2-4.4">When the source NVE receives the EVPN MAC/IP Advertisement route,
2635 it updates its IP-VRF table with the new adjacency information
2636 (pointing to the target NVE). In the case of the asymmetric IRB,
2637 the source NVE also updates its ARP table with the received
2638 adjacency information, and in the case of the symmetric IRB, the
2639 source NVE removes the entry associated with the received (IP, MAC) from its local ARP table. Furthermore, it withdraws its
2640 previously advertised EVPN MAC/IP route with both the MAC and IP
2641 address fields filled in.<a href="#section-7.2-4.4" class="pilcrow">¶</a>
2642 </li>
2643 <li class="normal" id="section-7.2-4.5">All other remote NVE devices, upon receiving the MAC/IP
2644 Advertisement route with the MAC Mobility extended community, compare
2645 the sequence number in this advertisement with the one previously
2646 received. If the new sequence number is greater than the old one,
2647 then they update the MAC/IP addresses of the TS in their
2648 corresponding MAC-VRF, IP-VRF, and ARP tables (in the case of
2649 asymmetric IRB) to point to the new NVE. Furthermore, upon
2650 receiving the MAC/IP withdraw for the TS from the old NVE, these
2651 remote PEs perform the cleanups for their BGP tables.<a href="#section-7.2-4.5" class="pilcrow">¶</a>
2652 </li>
2653 </ul>
2654 <p id="section-7.2-5">
2655 If an EVPN-IRB NVE is configured not to advertise MAC-only routes,
2656 then upon receiving the first data packet, it learns the MAC address
2657 of the TS and updates the MAC entry in the corresponding MAC-VRF
2658 table with the local adjacency information (e.g., local interface).
2659 It also realizes that there has been a MAC move because the same MAC
2660 address has been learned remotely from the source NVE. It uses the
2661 local MAC route to find the corresponding local MAC-IP route and
2662 sends a unicast ARP request to the host. When receiving an ARP
2663 response, it follows the procedure outlined in <a href="#sect-7.1" class="xref">Section 7.1</a>. In the
2664 prior case, where MAC-only routes are also advertised, this procedure
2665 of triggering a unicast ARP probe at the target PE <span class="bcp14">MAY</span> also be used
2666 in addition to the source PE broadcast ARP probing procedure
2667 described earlier for better convergence.<a href="#section-7.2-5" class="pilcrow">¶</a></p>
2668 </section>
2669 </div>
2670 <div id="sect-7.3">
2671 <section id="section-7.3">
2672 <h3 id="name-silent-host">
2673 <a href="#section-7.3" class="section-number selfRef">7.3. </a><a href="#name-silent-host" class="section-name selfRef">Silent Host</a>
2674 </h3>
2675 <p id="section-7.3-1">
2676 In this scenario, when a TS moves from a source NVE to a target NVE,
2677 the TS is silent, and it neither initiates an ARP request nor sends
2678 any data traffic. Therefore, neither the target nor the source NVEs
2679 are aware of the MAC move.<a href="#section-7.3-1" class="pilcrow">¶</a></p>
2680 <p id="section-7.3-2">
2681 On the source NVE, an age-out timer (for the silent host that has
2682 moved) is used to trigger an ARP probe. This age-out timer can be
2683 either an ARP timer or a MAC age-out timer, and this is an implementation
2684 choice. The ARP request gets sent both locally to all the attached
2685 TSs on that subnet as well as to all the remote NVEs
2686 (including the target NVE) participating in that subnet. The source
2687 NVE also withdraws the EVPN MAC/IP Advertisement route with only the
2688 MAC address (if it has previously advertised such a route).<a href="#section-7.3-2" class="pilcrow">¶</a></p>
2689 <p id="section-7.3-3">
2690 The target NVE passes the ARP request to its locally attached TSs, and when
2691 it receives the ARP response, it updates its MAC-VRF, IP-VRF, and ARP table
2692 with the host (IP, MAC) and local adjacency information (e.g., local
2693 interface). It also sends an EVPN MAC/IP Advertisement route with both the
2694 MAC and IP address fields filled in along with the MAC Mobility extended
2695 community, with the sequence number incremented by one.<a href="#section-7.3-3" class="pilcrow">¶</a></p>
2696 <p id="section-7.3-4">
2697 When the source NVE receives the EVPN MAC/IP Advertisement route, it
2698 updates its IP-VRF table with the new adjacency information (pointing
2699 to the target NVE). In the case of the asymmetric IRB, the source
2700 NVE also updates its ARP table with the received adjacency
2701 information, and in the case of the symmetric IRB, the source NVE
2702 removes the entry associated with the received (IP, MAC) from its
2703 local ARP table. Furthermore, it withdraws its previously advertised
2704 EVPN MAC/IP route with both the MAC and IP address fields filled in.<a href="#section-7.3-4" class="pilcrow">¶</a></p>
2705 <p id="section-7.3-5">
2706 All other remote NVE devices, upon receiving the MAC/IP Advertisement route
2707 with the MAC Mobility extended community, compare the sequence number in this
2708 advertisement with the one previously received. If the new sequence number
2709 is greater than the old one, then they update the MAC/IP addresses of the
2710 TS in their corresponding MAC-VRF, IP-VRF, and ARP (in the case of
2711 asymmetric IRB) tables to point to the new NVE. Furthermore, upon
2712 receiving the MAC/IP withdraw for the TS from the old NVE, these remote PEs
2713 perform the cleanups for their BGP tables.<a href="#section-7.3-5" class="pilcrow">¶</a></p>
2714 </section>
2715 </div>
2716 </section>
2717 </div>
2718 <div id="sect-8">
2719 <section id="section-8">
2720 <h2 id="name-bgp-encoding">
2721 <a href="#section-8" class="section-number selfRef">8. </a><a href="#name-bgp-encoding" class="section-name selfRef">BGP Encoding</a>
2722 </h2>
2723 <p id="section-8-1">
2724 This document defines one new BGP Extended Community for EVPN.<a href="#section-8-1" class="pilcrow">¶</a></p>
2725 <div id="sect-8.1">
2726 <section id="section-8.1">
2727 <h3 id="name-evpn-routers-mac-extended-c">
2728 <a href="#section-8.1" class="section-number selfRef">8.1. </a><a href="#name-evpn-routers-mac-extended-c" class="section-name selfRef">EVPN Router's MAC Extended Community</a>
2729 </h3>
2730 <p id="section-8.1-1">
2731 A new EVPN BGP Extended Community called "EVPN Router's MAC" is introduced
2732 here. This new extended community is a transitive extended community
2733 with a Type field of 0x06 (EVPN) and a Sub-Type field of 0x03. It may
2734 be advertised along with the Encapsulation Extended Community defined in
2735 <span><a href="https://www.rfc-editor.org/rfc/rfc9012#section-4.1" class="relref">Section 4.1</a> of [<a href="#RFC9012" class="xref">RFC9012</a>]</span>.<a href="#section-8.1-1" class="pilcrow">¶</a></p>
2736 <p id="section-8.1-2">
2737 The EVPN Router's MAC Extended Community is encoded as an 8-octet value as
2738 follows:<a href="#section-8.1-2" class="pilcrow">¶</a></p>
2739 <span id="name-evpn-routers-mac-extended-co"></span><div id="fig-5">
2740 <figure id="figure-5">
2741 <div class="alignLeft art-text artwork" id="section-8.1-3.1">
2742 <pre>
2743 0 1 2 3
2744 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
2745 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
2746 | Type=0x06 | Sub-Type=0x03 | EVPN Router's MAC |
2747 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
2748 | EVPN Router's MAC Cont'd |
2749 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
2750 </pre>
2751 </div>
2752 <figcaption><a href="#figure-5" class="selfRef">Figure 5</a>:
2753 <a href="#name-evpn-routers-mac-extended-co" class="selfRef">EVPN Router's MAC Extended Community</a>
2754 </figcaption></figure>
2755 </div>
2756 <p id="section-8.1-4">
2757 This extended community is used to carry the PE's MAC address for
2758 symmetric IRB scenarios, and it is sent with EVPN RT-2. The
2759 advertising PE <span class="bcp14">SHALL</span> only attach a single EVPN Router's MAC Extended
2760 Community to a route. In case the receiving PE receives more than
2761 one EVPN Router's MAC Extended Community with a route, it <span class="bcp14">SHALL</span> process
2762 the first one in the list and not store and propagate the others.<a href="#section-8.1-4" class="pilcrow">¶</a></p>
2763 </section>
2764 </div>
2765 </section>
2766 </div>
2767 <div id="sect-9">
2768 <section id="section-9">
2769 <h2 id="name-operational-models-for-symm">
2770 <a href="#section-9" class="section-number selfRef">9. </a><a href="#name-operational-models-for-symm" class="section-name selfRef">Operational Models for Symmetric Inter-Subnet Forwarding</a>
2771 </h2>
2772 <p id="section-9-1">
2773 The following sections describe two main symmetric IRB forwarding
2774 scenarios (within a DC -- i.e., intra-DC) along with the
2775 corresponding procedures. In the following scenarios, without loss
2776 of generality, it is assumed that a given tenant is represented by a
2777 single IP-VPN instance. Therefore, on a given PE, a tenant is
2778 represented by a single IP-VRF table and one or more MAC-VRF tables.<a href="#section-9-1" class="pilcrow">¶</a></p>
2779 <div id="sect-9.1">
2780 <section id="section-9.1">
2781 <h3 id="name-irb-forwarding-on-nves-for-">
2782 <a href="#section-9.1" class="section-number selfRef">9.1. </a><a href="#name-irb-forwarding-on-nves-for-" class="section-name selfRef">IRB Forwarding on NVEs for Tenant Systems</a>
2783 </h3>
2784 <p id="section-9.1-1">
2785 This section covers the symmetric IRB procedures for the scenario
2786 where each TS is attached to one or more NVEs, and its
2787 host IP and MAC addresses are learned by the attached NVEs and are
2788 distributed to all other NVEs that are interested in participating in
2789 both intra-subnet and inter-subnet communications with that TS.<a href="#section-9.1-1" class="pilcrow">¶</a></p>
2790 <p id="section-9.1-2">
2791 In this scenario, without loss of generality, it is assumed that NVEs
2792 operate in VLAN-based service interface mode with one bridge table(s)
2793 per MAC-VRF. Thus, for a given tenant, an NVE has one MAC-VRF for
2794 each tenant subnet (e.g., each VLAN) that is configured for extension
2795 via VXLAN or NVGRE encapsulation. In the case of VLAN-aware
2796 bundling, each MAC-VRF consists of multiple bridge tables (e.g.,
2797 one bridge table per VLAN). The MAC-VRFs on an NVE for a given
2798 tenant are associated with an IP-VRF corresponding to that tenant (or
2799 IP-VPN instance) via their IRB interfaces.<a href="#section-9.1-2" class="pilcrow">¶</a></p>
2800 <p id="section-9.1-3">
2801 Since VXLAN and NVGRE encapsulations require an inner Ethernet header
2802 (inner MAC SA/DA) and since a TS MAC address cannot be used for inter-subnet traffic, the ingress NVE's MAC address is used as an inner MAC
2803 SA. The NVE's MAC address is the device MAC address, and it is common
2804 across all MAC-VRFs and IP-VRFs. This MAC address is advertised
2805 using the new EVPN Router's MAC Extended Community (<a href="#sect-8.1" class="xref">Section 8.1</a>).<a href="#section-9.1-3" class="pilcrow">¶</a></p>
2806 <p id="section-9.1-4">
2807 <a href="#fig-6" class="xref">Figure 6</a> below illustrates this scenario, where a given tenant (e.g., an
2808 IP-VPN instance) has three subnets represented by MAC-VRF1, MAC-VRF2, and
2809 MAC-VRF3 across two NVEs. There are five TSs that are associated with
2810 these three MAC-VRFs -- i.e., TS1, TS4, and TS5 are on the same subnet
2811 (e.g., the same MAC-VRF/VLAN). TS1 and TS5 are associated with MAC-VRF1 on
2812 NVE1, while TS4 is associated with MAC-VRF1 on NVE2. TS2 is associated
2813 with MAC-VRF2 on NVE1, and TS3 is associated with MAC-VRF3 on NVE2.
2814 MAC-VRF1 and MAC-VRF2 on NVE1 are, in turn, associated with IP-VRF1 on NVE1,
2815 and MAC-VRF1 and MAC-VRF3 on NVE2 are associated with IP-VRF1 on NVE2.
2816 When TS1, TS5, and TS4 exchange traffic with each other, only the L2
2817 forwarding (bridging) part of the IRB solution is exercised because all
2818 these TSs belong to the same subnet. However, when TS1 wants to exchange
2819 traffic with TS2 or TS3, which belong to different subnets, both the bridging
2820 and routing parts of the IRB solution are exercised. The following
2821 subsections describe the control and data plane operations for this IRB
2822 scenario in detail.<a href="#section-9.1-4" class="pilcrow">¶</a></p>
2823 <span id="name-irb-forwarding-on-nves-for-t"></span><div id="fig-6">
2824 <figure id="figure-6">
2825 <div class="alignLeft art-text artwork" id="section-9.1-5.1">
2826 <pre>
2827 NVE1 +---------+
2828 +-------------+ | |
2829 TS1-----| MACx| | | NVE2
2830 (M1/IP1) |(MAC- | | | +-------------+
2831 TS5-----| VRF1)\ | | MPLS/ | |MACy (MAC- |-----TS3
2832 (M5/IP5) | \ | | VXLAN/ | | / VRF3) | (M3/IP3)
2833 | (IP-VRF1)|----| NVGRE |---|(IP-VRF1) |
2834 | / | | | | \ |
2835 TS2-----|(MAC- / | | | | (MAC- |-----TS4
2836 (M2/IP2) | VRF2) | | | | VRF1) | (M4/IP4)
2837 +-------------+ | | +-------------+
2838 | |
2839 +---------+
2840 </pre>
2841 </div>
2842 <figcaption><a href="#figure-6" class="selfRef">Figure 6</a>:
2843 <a href="#name-irb-forwarding-on-nves-for-t" class="selfRef">IRB Forwarding on NVEs for Tenant Systems</a>
2844 </figcaption></figure>
2845 </div>
2846 <div id="sect-9.1.1">
2847 <section id="section-9.1.1">
2848 <h4 id="name-control-plane-operation">
2849 <a href="#section-9.1.1" class="section-number selfRef">9.1.1. </a><a href="#name-control-plane-operation" class="section-name selfRef">Control Plane Operation</a>
2850 </h4>
2851 <p id="section-9.1.1-1">
2852 Each NVE advertises a MAC/IP Advertisement route (i.e., route type 2)
2853 for each of its TSs with the following field set:<a href="#section-9.1.1-1" class="pilcrow">¶</a></p>
2854 <ul class="normal">
2855 <li class="normal" id="section-9.1.1-2.1">RD and Ethernet Segment Identifier (ESI) per <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span><a href="#section-9.1.1-2.1" class="pilcrow">¶</a>
2856 </li>
2857 <li class="normal" id="section-9.1.1-2.2">Ethernet Tag = 0 (assuming VLAN-based service)<a href="#section-9.1.1-2.2" class="pilcrow">¶</a>
2858 </li>
2859 <li class="normal" id="section-9.1.1-2.3">MAC Address Length = 48<a href="#section-9.1.1-2.3" class="pilcrow">¶</a>
2860 </li>
2861 <li class="normal" id="section-9.1.1-2.4">MAC Address = Mi (where i = 1, 2, 3, 4, or 5) in <a href="#fig-6" class="xref">Figure 6</a>, above<a href="#section-9.1.1-2.4" class="pilcrow">¶</a>
2862 </li>
2863 <li class="normal" id="section-9.1.1-2.5">IP Address Length = 32 or 128<a href="#section-9.1.1-2.5" class="pilcrow">¶</a>
2864 </li>
2865 <li class="normal" id="section-9.1.1-2.6">IP Address = IPi (where i = 1, 2, 3, 4, or 5) in <a href="#fig-6" class="xref">Figure 6</a>, above<a href="#section-9.1.1-2.6" class="pilcrow">¶</a>
2866 </li>
2867 <li class="normal" id="section-9.1.1-2.7">Label1 = MPLS label or VNI corresponding to MAC-VRF<a href="#section-9.1.1-2.7" class="pilcrow">¶</a>
2868 </li>
2869 <li class="normal" id="section-9.1.1-2.8">Label2 = MPLS label or VNI corresponding to IP-VRF<a href="#section-9.1.1-2.8" class="pilcrow">¶</a>
2870 </li>
2871 </ul>
2872 <p id="section-9.1.1-3">
2873 Each NVE advertises an EVPN RT-2 route with two Route Targets (one
2874 corresponding to its MAC-VRF and the other corresponding to its IP-VRF).
2875 Furthermore, the EVPN RT-2 is advertised with two BGP Extended Communities.
2876 The first BGP Extended Community identifies the tunnel type, and it is
2877 called "Encapsulation Extended Community" as defined in
2878 <span>[<a href="#RFC9012" class="xref">RFC9012</a>]</span>, and the second BGP Extended Community includes
2879 the MAC address of the NVE (e.g., MACx for NVE1 or MACy for NVE2) as
2880 defined in <a href="#sect-8.1" class="xref">Section 8.1</a>. The EVPN Router's MAC Extended Community <span class="bcp14">MUST</span> be added
2881 when the Ethernet NVO tunnel is used. If the IP NVO tunnel type is used, then
2882 there is no need to send this second Extended Community. It should be
2883 noted that the IP NVO tunnel type is only applicable to symmetric IRB
2884 procedures.<a href="#section-9.1.1-3" class="pilcrow">¶</a></p>
2885 <p id="section-9.1.1-4">
2886 Upon receiving this advertisement, the receiving NVE performs the
2887 following:<a href="#section-9.1.1-4" class="pilcrow">¶</a></p>
2888 <ul class="normal">
2889 <li class="normal" id="section-9.1.1-5.1">It uses Route Targets corresponding to its MAC-VRF and IP-VRF for
2890 identifying these tables and subsequently importing the MAC and IP
2891 addresses into them, respectively.<a href="#section-9.1.1-5.1" class="pilcrow">¶</a>
2892 </li>
2893 <li class="normal" id="section-9.1.1-5.2">It imports the MAC address from the MAC/IP Advertisement route into
2894 the MAC-VRF with the BGP next-hop address as the underlay tunnel
2895 destination address (e.g., VTEP DA for VXLAN encapsulation) and
2896 label1 as the VNI for VXLAN encapsulation or an EVPN label for MPLS
2897 encapsulation.<a href="#section-9.1.1-5.2" class="pilcrow">¶</a>
2898 </li>
2899 <li class="normal" id="section-9.1.1-5.3">If the route carries the new EVPN Router's MAC Extended Community and
2900 if the receiving NVE uses an Ethernet NVO tunnel, then the receiving
2901 NVE imports the IP address into IP-VRF with NVE's MAC address
2902 (from the new EVPN Router's MAC Extended Community) as the inner MAC DA, the BGP next-hop address as the underlay tunnel destination address, the VTEP DA for VXLAN encapsulation, and label2 as the IP-VPN VNI for VXLAN
2903 encapsulation.<a href="#section-9.1.1-5.3" class="pilcrow">¶</a>
2904 </li>
2905 <li class="normal" id="section-9.1.1-5.4">If the receiving NVE uses MPLS encapsulation, then the receiving
2906 NVE imports the IP address into IP-VRF with the BGP next-hop address
2907 as the underlay tunnel destination address and label2 as the IP-VPN
2908 label for MPLS encapsulation.<a href="#section-9.1.1-5.4" class="pilcrow">¶</a>
2909 </li>
2910 </ul>
2911 <p id="section-9.1.1-6">
2912 If the receiving NVE receives an EVPN RT-2 with only label1 and only
2913 a single Route Target corresponding to IP-VRF; an
2914 EVPN RT-2 with only a single Route Target corresponding to MAC-VRF
2915 but with both label1 and label2; or an EVPN RT-2 with a
2916 MAC address length of zero, then it <span class="bcp14">MUST</span> use the treat-as-withdraw
2917 approach <span>[<a href="#RFC7606" class="xref">RFC7606</a>]</span> and <span class="bcp14">SHOULD</span> log an error message.<a href="#section-9.1.1-6" class="pilcrow">¶</a></p>
2918 </section>
2919 </div>
2920 <div id="sect-9.1.2">
2921 <section id="section-9.1.2">
2922 <h4 id="name-data-plane-operation">
2923 <a href="#section-9.1.2" class="section-number selfRef">9.1.2. </a><a href="#name-data-plane-operation" class="section-name selfRef">Data Plane Operation</a>
2924 </h4>
2925 <p id="section-9.1.2-1">
2926 The following description of the data plane operation describes just
2927 the logical functions, and the actual implementation may differ. Let's consider the data plane operation when TS1 in subnet-1 (MAC-VRF1) on NVE1
2928 wants to send traffic to TS3 in subnet-3 (MAC-VRF3) on NVE2.<a href="#section-9.1.2-1" class="pilcrow">¶</a></p>
2929 <ul class="normal">
2930 <li class="normal" id="section-9.1.2-2.1">NVE1 receives a packet with the MAC DA corresponding to the MAC-VRF1 IRB
2931 interface on NVE1 (the interface between MAC-VRF1 and IP-VRF1) and
2932 the VLAN tag corresponding to MAC-VRF1.<a href="#section-9.1.2-2.1" class="pilcrow">¶</a>
2933 </li>
2934 <li class="normal" id="section-9.1.2-2.2">Upon receiving the packet, the NVE1 uses the VLAN tag to identify the
2935 MAC-VRF1. It then looks up the MAC DA and forwards the frame to
2936 its IRB interface.<a href="#section-9.1.2-2.2" class="pilcrow">¶</a>
2937 </li>
2938 <li class="normal" id="section-9.1.2-2.3">The Ethernet header of the packet is stripped, and the packet is
2939 fed to the IP-VRF, where an IP lookup is performed on the
2940 destination IP address. NVE1 also decrements the TTL / hop limit
2941 for that packet by one, and if it reaches zero, NVE1 discards the
2942 packet. This lookup yields the outgoing NVO tunnel and the
2943 required encapsulation. If the encapsulation is for the Ethernet NVO
2944 tunnel, then it includes the egress NVE's MAC address as the inner MAC
2945 DA, the egress NVE's IP address (e.g., BGP next-hop address) as
2946 the VTEP DA, and the VPN-ID as the VNI. The inner MAC SA and VTEP
2947 SA are set to NVE's MAC and IP addresses, respectively. If it is an
2948 MPLS encapsulation, then the corresponding EVPN and LSP labels are
2949 added to the packet. The packet is then forwarded to the egress
2950 NVE.<a href="#section-9.1.2-2.3" class="pilcrow">¶</a>
2951 </li>
2952 <li class="normal" id="section-9.1.2-2.4">If the egress NVE receives a packet from the Ethernet NVO tunnel (e.g., it is VXLAN encapsulated),
2953 then it removes the Ethernet header. Since the inner MAC DA is the egress NVE's MAC address,
2954 the egress NVE knows that it needs to perform an IP lookup. It
2955 uses the VNI to identify the IP-VRF table. If the packet is MPLS
2956 encapsulated, then the EVPN label lookup identifies the IP-VRF
2957 table. Next, an IP lookup is performed for the destination TS
2958 (TS3), which results in an access-facing IRB interface over which
2959 the packet is sent. Before sending the packet over this
2960 interface, the ARP table is consulted to get the destination TS's
2961 MAC address. NVE2 also decrements the TTL / hop limit for that
2962 packet by one, and if it reaches zero, NVE2 discards the packet.<a href="#section-9.1.2-2.4" class="pilcrow">¶</a>
2963 </li>
2964 <li class="normal" id="section-9.1.2-2.5">The IP packet is encapsulated with an Ethernet header, with the MAC SA
2965 set to that of the IRB interface MAC address (i.e., the IRB interface
2966 between MAC-VRF3 and IP-VRF1 on NVE2) and the MAC DA set to that of the
2967 destination TS (TS3) MAC address. The packet is sent to the
2968 corresponding MAC-VRF (i.e., MAC-VRF3) and, after a lookup of MAC
2969 DA, is forwarded to the destination TS (TS3) over the
2970 corresponding interface.<a href="#section-9.1.2-2.5" class="pilcrow">¶</a>
2971 </li>
2972 </ul>
2973 <p id="section-9.1.2-3">
2974 In this symmetric IRB scenario, inter-subnet traffic between NVEs
2975 will always use the IP-VRF VNI/MPLS label. For instance, traffic
2976 from TS2 to TS4 will be encapsulated by NVE1 using NVE2's IP-VRF VNI/MPLS label, as long as TS4's host IP is present in NVE1's IP-VRF.<a href="#section-9.1.2-3" class="pilcrow">¶</a></p>
2977 </section>
2978 </div>
2979 </section>
2980 </div>
2981 <div id="sect-9.2">
2982 <section id="section-9.2">
2983 <h3 id="name-irb-forwarding-on-nves-for-s">
2984 <a href="#section-9.2" class="section-number selfRef">9.2. </a><a href="#name-irb-forwarding-on-nves-for-s" class="section-name selfRef">IRB Forwarding on NVEs for Subnets behind Tenant Systems</a>
2985 </h3>
2986 <p id="section-9.2-1">
2987 This section covers the symmetric IRB procedures for the scenario where
2988 some TSs support one or more subnets and these TSs are
2989 associated with one or more NVEs. Therefore, besides the advertisement of
2990 MAC/IP addresses for each TS, which can be multihomed with All-Active
2991 redundancy mode, the associated NVE needs to also advertise the subnets
2992 statically configured on each TS.<a href="#section-9.2-1" class="pilcrow">¶</a></p>
2993 <p id="section-9.2-2">
2994 The main difference between this solution and the previous one is the
2995 additional advertisement corresponding to each subnet. These subnet
2996 advertisements are accomplished using the EVPN IP Prefix route
2997 defined in <span>[<a href="#RFC9136" class="xref">RFC9136</a>]</span>. These subnet
2998 prefixes are advertised with the IP address of their associated TS
2999 (which is in an overlay address space) as their next hop. The receiving
3000 NVEs perform recursive route resolution to resolve the subnet prefix
3001 with its advertising NVE so that they know which NVE to forward the
3002 packets to when they are destined for that subnet prefix.<a href="#section-9.2-2" class="pilcrow">¶</a></p>
3003 <p id="section-9.2-3">
3004 The advantage of this recursive route resolution is that when a TS
3005 moves from one NVE to another, there is no need to re-advertise any
3006 of the subnet prefixes for that TS. All that is needed is to advertise
3007 the IP/MAC addresses associated with the TS itself and exercise the MAC
3008 Mobility procedures for that TS. The recursive route resolution
3009 automatically takes care of the updates for the subnet prefixes of
3010 that TS.<a href="#section-9.2-3" class="pilcrow">¶</a></p>
3011 <p id="section-9.2-4">
3012 <a href="#fig-7" class="xref">Figure 7</a> illustrates this scenario where a given tenant (e.g., an IP-VPN
3013 service) has three subnets represented by MAC-VRF1, MAC-VRF2, and MAC-VRF3
3014 across two NVEs. There are four TSs associated with these three MAC-VRFs
3015 -- i.e., TS1 is connected to MAC-VRF1 on NVE1, TS2 is connected to MAC-VRF2
3016 on NVE1, TS3 is connected to MAC-VRF3 on NVE2, and TS4 is connected to
3017 MAC-VRF1 on NVE2. TS1 has two subnet prefixes (SN1 and SN2), and TS3 has a
3018 single subnet prefix (SN3). The MAC-VRFs on each NVE are associated with
3019 their corresponding IP-VRF using their IRB interfaces. When TS4 and TS1
3020 exchange intra-subnet traffic, only the L2 forwarding (bridging) part of the
3021 IRB solution is used (i.e., the traffic only goes through their MAC-VRFs);
3022 however, when TS3 wants to forward traffic to SN1 or SN2 sitting behind TS1
3023 (inter-subnet traffic), then both the bridging and routing parts of the IRB
3024 solution are exercised (i.e., the traffic goes through the corresponding
3025 MAC-VRFs and IP-VRFs).
3026
3027 If TS4, for example, wants to reach SN1, it uses
3028 its default route and sends the packet to the MAC address associated with
3029 the IRB interface on NVE2; NVE2 then performs an IP lookup in its IP-VRF and
3030 finds an entry for SN1. The following subsections describe the control and
3031 data plane operations for this IRB scenario in detail.<a href="#section-9.2-4" class="pilcrow">¶</a></p>
3032 <span id="name-irb-forwarding-on-nves-for-su"></span><div id="fig-7">
3033 <figure id="figure-7">
3034 <div class="alignLeft art-text artwork" id="section-9.2-5.1">
3035 <pre>
3036 NVE1 +----------+
3037 SN1--+ +-------------+ | |
3038 |--TS1-----|(MAC- \ | | |
3039 SN2--+ M1/IP1 | VRF1) \ | | |
3040 | (IP-VRF)|---| |
3041 | / | | |
3042 TS2-----|(MAC- / | | MPLS/ |
3043 M2/IP2 | VRF2) | | VXLAN/ |
3044 +-------------+ | NVGRE |
3045 +-------------+ | |
3046 SN3--+--TS3-----|(MAC-\ | | |
3047 M3/IP3 | VRF3)\ | | |
3048 | (IP-VRF)|---| |
3049 | / | | |
3050 TS4-----|(MAC- / | | |
3051 M4/IP4 | VRF1) | | |
3052 +-------------+ +----------+
3053 NVE2
3054 </pre>
3055 </div>
3056 <figcaption><a href="#figure-7" class="selfRef">Figure 7</a>:
3057 <a href="#name-irb-forwarding-on-nves-for-su" class="selfRef">IRB Forwarding on NVEs for Subnets behind TSs</a>
3058 </figcaption></figure>
3059 </div>
3060 <p id="section-9.2-6">
3061 Note that in <a href="#fig-7" class="xref">Figure 7</a>, above, SN1 and SN2 are configured on NVE1,
3062 which then advertises each in an IP Prefix route. Similarly, SN3 is
3063 configured on NVE2, which then advertises it in an IP Prefix route.<a href="#section-9.2-6" class="pilcrow">¶</a></p>
3064 <div id="sect-9.2.1">
3065 <section id="section-9.2.1">
3066 <h4 id="name-control-plane-operation-2">
3067 <a href="#section-9.2.1" class="section-number selfRef">9.2.1. </a><a href="#name-control-plane-operation-2" class="section-name selfRef">Control Plane Operation</a>
3068 </h4>
3069 <p id="section-9.2.1-1">
3070 Each NVE advertises a route type 5 (EVPN RT-5, IP Prefix route
3071 defined in <span>[<a href="#RFC9136" class="xref">RFC9136</a>]</span>) for each of its
3072 subnet prefixes with the IP address of its TS as the next hop
3073 (Gateway Address field) as follows:<a href="#section-9.2.1-1" class="pilcrow">¶</a></p>
3074 <ul class="normal">
3075 <li class="normal" id="section-9.2.1-2.1">RD associated with the IP-VRF<a href="#section-9.2.1-2.1" class="pilcrow">¶</a>
3076 </li>
3077 <li class="normal" id="section-9.2.1-2.2">ESI = 0<a href="#section-9.2.1-2.2" class="pilcrow">¶</a>
3078 </li>
3079 <li class="normal" id="section-9.2.1-2.3">Ethernet Tag = 0<a href="#section-9.2.1-2.3" class="pilcrow">¶</a>
3080 </li>
3081 <li class="normal" id="section-9.2.1-2.4">IP Prefix Length = 0 to 32 or 0 to 128<a href="#section-9.2.1-2.4" class="pilcrow">¶</a>
3082 </li>
3083 <li class="normal" id="section-9.2.1-2.5">IP Prefix = SNi<a href="#section-9.2.1-2.5" class="pilcrow">¶</a>
3084 </li>
3085 <li class="normal" id="section-9.2.1-2.6">Gateway Address = IPi (IP address of TS)<a href="#section-9.2.1-2.6" class="pilcrow">¶</a>
3086 </li>
3087 <li class="normal" id="section-9.2.1-2.7">MPLS Label = 0<a href="#section-9.2.1-2.7" class="pilcrow">¶</a>
3088 </li>
3089 </ul>
3090 <p id="section-9.2.1-3">
3091 This EVPN RT-5 is advertised with one or more Route Targets associated with
3092 the IP-VRF from which the route is originated.<a href="#section-9.2.1-3" class="pilcrow">¶</a></p>
3093 <p id="section-9.2.1-4">
3094
3095 Each NVE also advertises an EVPN RT-2 (MAC/IP Advertisement route)
3096 along with its associated Route Targets and Extended Communities
3097 for each of its TSs exactly as described in <a href="#sect-9.1.1" class="xref">Section 9.1.1</a>.<a href="#section-9.2.1-4" class="pilcrow">¶</a></p>
3098 <p id="section-9.2.1-5">
3099 Upon receiving the EVPN RT-5 advertisement, the receiving NVE
3100 performs the following:<a href="#section-9.2.1-5" class="pilcrow">¶</a></p>
3101 <ul class="normal">
3102 <li class="normal" id="section-9.2.1-6.1">It uses the Route Target to identify the corresponding IP-VRF.<a href="#section-9.2.1-6.1" class="pilcrow">¶</a>
3103 </li>
3104 <li class="normal" id="section-9.2.1-6.2">It imports the IP prefix into its corresponding IP-VRF
3105 configured with an import RT that is one of the RTs being carried
3106 by the EVPN RT-5 route, along with the IP address of the associated
3107 TS as its next hop.<a href="#section-9.2.1-6.2" class="pilcrow">¶</a>
3108 </li>
3109 </ul>
3110 <p id="section-9.2.1-7">
3111 When receiving the EVPN RT-2 advertisement, the receiving NVE imports the
3112 MAC/IP addresses of the TS into the corresponding MAC-VRF and IP-VRF
3113 per <a href="#sect-9.1.1" class="xref">Section 9.1.1</a>. When both routes exist, recursive route
3114 resolution is performed to resolve the IP prefix (received in EVPN
3115 RT-5) to its corresponding NVE's IP address (e.g., its BGP next hop).
3116 The BGP next hop will be used as the underlay tunnel destination address
3117 (e.g., VTEP DA for VXLAN encapsulation), and the EVPN Router's MAC will be used
3118 as the inner MAC for VXLAN encapsulation.<a href="#section-9.2.1-7" class="pilcrow">¶</a></p>
3119 </section>
3120 </div>
3121 <div id="sect-9.2.2">
3122 <section id="section-9.2.2">
3123 <h4 id="name-data-plane-operation-2">
3124 <a href="#section-9.2.2" class="section-number selfRef">9.2.2. </a><a href="#name-data-plane-operation-2" class="section-name selfRef">Data Plane Operation</a>
3125 </h4>
3126 <p id="section-9.2.2-1">
3127 The following description of the data plane operation describes just
3128 the logical functions, and the actual implementation may differ. Let's consider the data plane operation when a host in SN1 behind TS1 wants to send traffic
3129 to a host in SN3 behind TS3.<a href="#section-9.2.2-1" class="pilcrow">¶</a></p>
3130 <ul class="normal">
3131 <li class="normal" id="section-9.2.2-2.1">TS1 sends a packet with MAC DA corresponding to the MAC-VRF1 IRB
3132 interface of NVE1 and a VLAN tag corresponding to MAC-VRF1.<a href="#section-9.2.2-2.1" class="pilcrow">¶</a>
3133 </li>
3134 <li class="normal" id="section-9.2.2-2.2">Upon receiving the packet, the ingress NVE1 uses the VLAN tag to
3135 identify the MAC-VRF1. It then looks up the MAC DA and forwards
3136 the frame to its IRB interface as in <a href="#sect-9.1.1" class="xref">Section 9.1.1</a>.<a href="#section-9.2.2-2.2" class="pilcrow">¶</a>
3137 </li>
3138 <li class="normal" id="section-9.2.2-2.3">The Ethernet header of the packet is stripped, and the packet is
3139 fed to the IP-VRF, where an IP lookup is performed on the
3140 destination address.
3141
3142 This lookup yields the fields needed for
3143 VXLAN encapsulation with NVE2's MAC address as the inner MAC DA,
3144 NVE2's IP address as the VTEP DA, and the VNI. The MAC SA is set to
3145 NVE1's MAC address, and the VTEP SA is set to NVE1's IP address. NVE1
3146 also decrements the TTL / hop limit for that packet by one, and if it
3147 reaches zero, NVE1 discards the packet.<a href="#section-9.2.2-2.3" class="pilcrow">¶</a>
3148 </li>
3149 <li class="normal" id="section-9.2.2-2.4">The packet is then encapsulated with the proper header based on
3150 the above info and is forwarded to the egress NVE (NVE2).<a href="#section-9.2.2-2.4" class="pilcrow">¶</a>
3151 </li>
3152 <li class="normal" id="section-9.2.2-2.5">On the egress NVE (NVE2), assuming the packet is VXLAN
3153 encapsulated, the VXLAN and the inner Ethernet headers are removed,
3154 and the resultant IP packet is fed to the IP-VRF associated with
3155 that VNI.<a href="#section-9.2.2-2.5" class="pilcrow">¶</a>
3156 </li>
3157 <li class="normal" id="section-9.2.2-2.6">Next, a lookup is performed based on the IP DA (which is in SN3) in the
3158 associated IP-VRF of NVE2. The IP lookup yields the access-facing IRB
3159 interface over which the packet needs to be sent. Before sending the
3160 packet over this interface, the ARP table is consulted to get the
3161 destination TS (TS3) MAC address. NVE2 also decrements the TTL / hop
3162 limit for that packet by one, and if it reaches zero, NVE2 discards the
3163 packet.<a href="#section-9.2.2-2.6" class="pilcrow">¶</a>
3164 </li>
3165 <li class="normal" id="section-9.2.2-2.7">The IP packet is encapsulated with an Ethernet header with the MAC
3166 SA set to that of the access-facing IRB interface of the egress
3167 NVE (NVE2), and the MAC DA is set to that of the destination TS (TS3)
3168 MAC address. The packet is sent to the corresponding MAC-VRF3 and,
3169 after a lookup of MAC DA, is forwarded to the destination TS (TS3)
3170 over the corresponding interface.<a href="#section-9.2.2-2.7" class="pilcrow">¶</a>
3171 </li>
3172 </ul>
3173 </section>
3174 </div>
3175 </section>
3176 </div>
3177 </section>
3178 </div>
3179 <div id="sect-11">
3180 <section id="section-10">
3181 <h2 id="name-security-considerations">
3182 <a href="#section-10" class="section-number selfRef">10. </a><a href="#name-security-considerations" class="section-name selfRef">Security Considerations</a>
3183 </h2>
3184 <p id="section-10-1">
3185 The security considerations for Layer 2 forwarding in this document
3186 follow those of <span>[<a href="#RFC7432" class="xref">RFC7432</a>]</span> for MPLS encapsulation and those
3187 of <span>[<a href="#RFC8365" class="xref">RFC8365</a>]</span> for VXLAN or NVGRE encapsulations. This section
3188 describes additional considerations.<a href="#section-10-1" class="pilcrow">¶</a></p>
3189 <p id="section-10-2">
3190 This document describes a set of procedures for inter-subnet
3191 forwarding of tenant traffic across PEs (or NVEs). These procedures
3192 include both Layer 2 forwarding and Layer 3 routing on a packet-by-packet basis. The security consideration for Layer 3 routing in this
3193 document follows that of <span>[<a href="#RFC4365" class="xref">RFC4365</a>]</span>, with the exception of the
3194 application of routing protocols between CEs and PEs. Contrary to
3195 <span>[<a href="#RFC4364" class="xref">RFC4364</a>]</span>, this document does not describe route distribution
3196 techniques between CEs and PEs but rather considers the CEs as TSs
3197 or VAs that do not run dynamic routing protocols. This can be
3198 considered a security advantage, since dynamic routing protocols can
3199 be blocked on the NVE/PE ACs, not allowing the tenant to interact
3200 with the infrastructure's dynamic routing protocols.<a href="#section-10-2" class="pilcrow">¶</a></p>
3201 <p id="section-10-3">
3202 The VPN scheme described in this document does not provide the
3203 quartet of security properties mentioned in <span>[<a href="#RFC4365" class="xref">RFC4365</a>]</span>
3204 (confidentiality protection, source authentication, integrity
3205 protection, and replay protection). If these are desired, they must be
3206 provided by mechanisms that are outside the scope of the VPN
3207 mechanisms.<a href="#section-10-3" class="pilcrow">¶</a></p>
3208 <p id="section-10-4">
3209 In this document, the EVPN RT-5 is used for certain scenarios. This
3210 route uses an Overlay Index that requires a recursive resolution to a
3211 different EVPN route (an EVPN RT-2). Because of this, it is worth
3212 noting that any action that ends up filtering or modifying the EVPN
3213 RT-2 route used to convey the Overlay Indexes will modify the
3214 resolution of the EVPN RT-5 and therefore the forwarding of packets
3215 to the remote subnet.<a href="#section-10-4" class="pilcrow">¶</a></p>
3216 </section>
3217 </div>
3218 <div id="sect-12">
3219 <section id="section-11">
3220 <h2 id="name-iana-considerations">
3221 <a href="#section-11" class="section-number selfRef">11. </a><a href="#name-iana-considerations" class="section-name selfRef">IANA Considerations</a>
3222 </h2>
3223 <p id="section-11-1">
3224 IANA has allocated Sub-Type value 0x03 in the "EVPN Extended Community Sub-Types" registry as follows:<a href="#section-11-1" class="pilcrow">¶</a></p>
3225 <div id="IANA_table">
3226 <table class="center" id="table-1">
3227 <caption><a href="#table-1" class="selfRef">Table 1</a></caption>
3228 <thead>
3229 <tr>
3230 <th class="text-left" rowspan="1" colspan="1">Sub-Type Value</th>
3231 <th class="text-left" rowspan="1" colspan="1">Name</th>
3232 <th class="text-left" rowspan="1" colspan="1">Reference</th>
3233 </tr>
3234 </thead>
3235 <tbody>
3236 <tr>
3237 <td class="text-left" rowspan="1" colspan="1">0x03</td>
3238 <td class="text-left" rowspan="1" colspan="1">EVPN Router's MAC Extended Community</td>
3239 <td class="text-left" rowspan="1" colspan="1">RFC 9135</td>
3240 </tr>
3241 </tbody>
3242 </table>
3243 </div>
3244 <p id="section-11-3">
3245 This document has been listed as an additional reference for the MAC/IP Advertisement route in the "EVPN Route Types" registry.<a href="#section-11-3" class="pilcrow">¶</a></p>
3246 </section>
3247 </div>
3248 <section id="section-12">
3249 <h2 id="name-references">
3250 <a href="#section-12" class="section-number selfRef">12. </a><a href="#name-references" class="section-name selfRef">References</a>
3251 </h2>
3252 <section id="section-12.1">
3253 <h3 id="name-normative-references">
3254 <a href="#section-12.1" class="section-number selfRef">12.1. </a><a href="#name-normative-references" class="section-name selfRef">Normative References</a>
3255 </h3>
3256 <dl class="references">
3257 <dt id="RFC2119">[RFC2119]</dt>
3258 <dd>
3259 <span class="refAuthor">Bradner, S.</span>, <span class="refTitle">"Key words for use in RFCs to Indicate Requirement Levels"</span>, <span class="seriesInfo">BCP 14</span>, <span class="seriesInfo">RFC 2119</span>, <span class="seriesInfo">DOI 10.17487/RFC2119</span>, <time datetime="1997-03" class="refDate">March 1997</time>, <span>&lt;<a href="https://www.rfc-editor.org/info/rfc2119">https://www.rfc-editor.org/info/rfc2119</a>&gt;</span>. </dd>
3260 <dd class="break"></dd>
3261 <dt id="RFC4364">[RFC4364]</dt>
3262 <dd>
3263 <span class="refAuthor">Rosen, E.</span> and <span class="refAuthor">Y. Rekhter</span>, <span class="refTitle">"BGP/MPLS IP Virtual Private Networks (VPNs)"</span>, <span class="seriesInfo">RFC 4364</span>, <span class="seriesInfo">DOI 10.17487/RFC4364</span>, <time datetime="2006-02" class="refDate">February 2006</time>, <span>&lt;<a href="https://www.rfc-editor.org/info/rfc4364">https://www.rfc-editor.org/info/rfc4364</a>&gt;</span>. </dd>
3264 <dd class="break"></dd>
3265 <dt id="RFC7432">[RFC7432]</dt>
3266 <dd>
3267 <span class="refAuthor">Sajassi, A., Ed.</span>, <span class="refAuthor">Aggarwal, R.</span>, <span class="refAuthor">Bitar, N.</span>, <span class="refAuthor">Isaac, A.</span>, <span class="refAuthor">Uttaro, J.</span>, <span class="refAuthor">Drake, J.</span>, and <span class="refAuthor">W. Henderickx</span>, <span class="refTitle">"BGP MPLS-Based Ethernet VPN"</span>, <span class="seriesInfo">RFC 7432</span>, <span class="seriesInfo">DOI 10.17487/RFC7432</span>, <time datetime="2015-02" class="refDate">February 2015</time>, <span>&lt;<a href="https://www.rfc-editor.org/info/rfc7432">https://www.rfc-editor.org/info/rfc7432</a>&gt;</span>. </dd>
3268 <dd class="break"></dd>
3269 <dt id="RFC7606">[RFC7606]</dt>
3270 <dd>
3271 <span class="refAuthor">Chen, E., Ed.</span>, <span class="refAuthor">Scudder, J., Ed.</span>, <span class="refAuthor">Mohapatra, P.</span>, and <span class="refAuthor">K. Patel</span>, <span class="refTitle">"Revised Error Handling for BGP UPDATE Messages"</span>, <span class="seriesInfo">RFC 7606</span>, <span class="seriesInfo">DOI 10.17487/RFC7606</span>, <time datetime="2015-08" class="refDate">August 2015</time>, <span>&lt;<a href="https://www.rfc-editor.org/info/rfc7606">https://www.rfc-editor.org/info/rfc7606</a>&gt;</span>. </dd>
3272 <dd class="break"></dd>
3273 <dt id="RFC8174">[RFC8174]</dt>
3274 <dd>
3275 <span class="refAuthor">Leiba, B.</span>, <span class="refTitle">"Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words"</span>, <span class="seriesInfo">BCP 14</span>, <span class="seriesInfo">RFC 8174</span>, <span class="seriesInfo">DOI 10.17487/RFC8174</span>, <time datetime="2017-05" class="refDate">May 2017</time>, <span>&lt;<a href="https://www.rfc-editor.org/info/rfc8174">https://www.rfc-editor.org/info/rfc8174</a>&gt;</span>. </dd>
3276 <dd class="break"></dd>
3277 <dt id="RFC8365">[RFC8365]</dt>
3278 <dd>
3279 <span class="refAuthor">Sajassi, A., Ed.</span>, <span class="refAuthor">Drake, J., Ed.</span>, <span class="refAuthor">Bitar, N.</span>, <span class="refAuthor">Shekhar, R.</span>, <span class="refAuthor">Uttaro, J.</span>, and <span class="refAuthor">W. Henderickx</span>, <span class="refTitle">"A Network Virtualization Overlay Solution Using Ethernet VPN (EVPN)"</span>, <span class="seriesInfo">RFC 8365</span>, <span class="seriesInfo">DOI 10.17487/RFC8365</span>, <time datetime="2018-03" class="refDate">March 2018</time>, <span>&lt;<a href="https://www.rfc-editor.org/info/rfc8365">https://www.rfc-editor.org/info/rfc8365</a>&gt;</span>. </dd>
3280 <dd class="break"></dd>
3281 <dt id="RFC9012">[RFC9012]</dt>
3282 <dd>
3283 <span class="refAuthor">Patel, K.</span>, <span class="refAuthor">Van de Velde, G.</span>, <span class="refAuthor">Sangli, S.</span>, and <span class="refAuthor">J. Scudder</span>, <span class="refTitle">"The BGP Tunnel Encapsulation Attribute"</span>, <span class="seriesInfo">RFC 9012</span>, <span class="seriesInfo">DOI 10.17487/RFC9012</span>, <time datetime="2021-04" class="refDate">April 2021</time>, <span>&lt;<a href="https://www.rfc-editor.org/info/rfc9012">https://www.rfc-editor.org/info/rfc9012</a>&gt;</span>. </dd>
3284 <dd class="break"></dd>
3285 <dt id="RFC9136">[RFC9136]</dt>
3286 <dd>
3287 <span class="refAuthor">Rabadan, J., Ed.</span>, <span class="refAuthor">Henderickx, W.</span>, <span class="refAuthor">Drake, J.</span>, <span class="refAuthor">Lin, W.</span>, and <span class="refAuthor">A. Sajassi</span>, <span class="refTitle">"IP Prefix Advertisement in Ethernet VPN (EVPN)"</span>, <span class="seriesInfo">RFC 9136</span>, <span class="seriesInfo">DOI 10.17487/RFC9136</span>, <time datetime="2021-10" class="refDate">October 2021</time>, <span>&lt;<a href="https://www.rfc-editor.org/info/rfc9136">https://www.rfc-editor.org/info/rfc9136</a>&gt;</span>. </dd>
3288 <dd class="break"></dd>
3289 </dl>
3290 </section>
3291 <section id="section-12.2">
3292 <h3 id="name-informative-references">
3293 <a href="#section-12.2" class="section-number selfRef">12.2. </a><a href="#name-informative-references" class="section-name selfRef">Informative References</a>
3294 </h3>
3295 <dl class="references">
3296 <dt id="I-D.ietf-bess-evpn-modes-interop">[EVPN]</dt>
3297 <dd>
3298 <span class="refAuthor">Krattiger, L., Ed.</span>, <span class="refAuthor">Sajassi, A., Ed.</span>, <span class="refAuthor">Thoria, S.</span>, <span class="refAuthor">Rabadan, J.</span>, and <span class="refAuthor">J. Drake</span>, <span class="refTitle">"EVPN Interoperability Modes"</span>, <span class="refContent">Work in Progress</span>, <span class="seriesInfo">Internet-Draft, draft-ietf-bess-evpn-modes-interop-00</span>, <time datetime="2021-05-26" class="refDate">26 May 2021</time>, <span>&lt;<a href="https://datatracker.ietf.org/doc/html/draft-ietf-bess-evpn-modes-interop-00">https://datatracker.ietf.org/doc/html/draft-ietf-bess-evpn-modes-interop-00</a>&gt;</span>. </dd>
3299 <dd class="break"></dd>
3300 <dt id="I-D.ietf-bess-evpn-irb-extended-mobility">[EXTENDED-MOBILITY]</dt>
3301 <dd>
3302 <span class="refAuthor">Malhotra, N., Ed.</span>, <span class="refAuthor">Sajassi, A.</span>, <span class="refAuthor">Pattekar, A.</span>, <span class="refAuthor">Rabadan, J.</span>, <span class="refAuthor">Lingala, A.</span>, and <span class="refAuthor">J. Drake</span>, <span class="refTitle">"Extended Mobility Procedures for EVPN-IRB"</span>, <span class="refContent">Work in Progress</span>, <span class="seriesInfo">Internet-Draft, draft-ietf-bess-evpn-irb-extended-mobility-07</span>, <time datetime="2021-10-02" class="refDate">2 October 2021</time>, <span>&lt;<a href="https://datatracker.ietf.org/doc/html/draft-ietf-bess-evpn-irb-extended-mobility-07">https://datatracker.ietf.org/doc/html/draft-ietf-bess-evpn-irb-extended-mobility-07</a>&gt;</span>. </dd>
3303 <dd class="break"></dd>
3304 <dt id="RFC4365">[RFC4365]</dt>
3305 <dd>
3306 <span class="refAuthor">Rosen, E.</span>, <span class="refTitle">"Applicability Statement for BGP/MPLS IP Virtual Private Networks (VPNs)"</span>, <span class="seriesInfo">RFC 4365</span>, <span class="seriesInfo">DOI 10.17487/RFC4365</span>, <time datetime="2006-02" class="refDate">February 2006</time>, <span>&lt;<a href="https://www.rfc-editor.org/info/rfc4365">https://www.rfc-editor.org/info/rfc4365</a>&gt;</span>. </dd>
3307 <dd class="break"></dd>
3308 <dt id="RFC5798">[RFC5798]</dt>
3309 <dd>
3310 <span class="refAuthor">Nadas, S., Ed.</span>, <span class="refTitle">"Virtual Router Redundancy Protocol (VRRP) Version 3 for IPv4 and IPv6"</span>, <span class="seriesInfo">RFC 5798</span>, <span class="seriesInfo">DOI 10.17487/RFC5798</span>, <time datetime="2010-03" class="refDate">March 2010</time>, <span>&lt;<a href="https://www.rfc-editor.org/info/rfc5798">https://www.rfc-editor.org/info/rfc5798</a>&gt;</span>. </dd>
3311 <dd class="break"></dd>
3312 <dt id="RFC7348">[RFC7348]</dt>
3313 <dd>
3314 <span class="refAuthor">Mahalingam, M.</span>, <span class="refAuthor">Dutt, D.</span>, <span class="refAuthor">Duda, K.</span>, <span class="refAuthor">Agarwal, P.</span>, <span class="refAuthor">Kreeger, L.</span>, <span class="refAuthor">Sridhar, T.</span>, <span class="refAuthor">Bursell, M.</span>, and <span class="refAuthor">C. Wright</span>, <span class="refTitle">"Virtual eXtensible Local Area Network (VXLAN): A Framework for Overlaying Virtualized Layer 2 Networks over Layer 3 Networks"</span>, <span class="seriesInfo">RFC 7348</span>, <span class="seriesInfo">DOI 10.17487/RFC7348</span>, <time datetime="2014-08" class="refDate">August 2014</time>, <span>&lt;<a href="https://www.rfc-editor.org/info/rfc7348">https://www.rfc-editor.org/info/rfc7348</a>&gt;</span>. </dd>
3315 <dd class="break"></dd>
3316 <dt id="RFC7365">[RFC7365]</dt>
3317 <dd>
3318 <span class="refAuthor">Lasserre, M.</span>, <span class="refAuthor">Balus, F.</span>, <span class="refAuthor">Morin, T.</span>, <span class="refAuthor">Bitar, N.</span>, and <span class="refAuthor">Y. Rekhter</span>, <span class="refTitle">"Framework for Data Center (DC) Network Virtualization"</span>, <span class="seriesInfo">RFC 7365</span>, <span class="seriesInfo">DOI 10.17487/RFC7365</span>, <time datetime="2014-10" class="refDate">October 2014</time>, <span>&lt;<a href="https://www.rfc-editor.org/info/rfc7365">https://www.rfc-editor.org/info/rfc7365</a>&gt;</span>. </dd>
3319 <dd class="break"></dd>
3320 <dt id="RFC7637">[RFC7637]</dt>
3321 <dd>
3322 <span class="refAuthor">Garg, P., Ed.</span> and <span class="refAuthor">Y. Wang, Ed.</span>, <span class="refTitle">"NVGRE: Network Virtualization Using Generic Routing Encapsulation"</span>, <span class="seriesInfo">RFC 7637</span>, <span class="seriesInfo">DOI 10.17487/RFC7637</span>, <time datetime="2015-09" class="refDate">September 2015</time>, <span>&lt;<a href="https://www.rfc-editor.org/info/rfc7637">https://www.rfc-editor.org/info/rfc7637</a>&gt;</span>. </dd>
3323 <dd class="break"></dd>
3324 <dt id="I-D.ietf-nvo3-vxlan-gpe">[VXLAN-GPE]</dt>
3325 <dd>
3326 <span class="refAuthor">Maino, F., Ed.</span>, <span class="refAuthor">Kreeger, L., Ed.</span>, and <span class="refAuthor">U. Elzur, Ed.</span>, <span class="refTitle">"Generic Protocol Extension for VXLAN (VXLAN-GPE)"</span>, <span class="refContent">Work in Progress</span>, <span class="seriesInfo">Internet-Draft, draft-ietf-nvo3-vxlan-gpe-12</span>, <time datetime="2021-09-22" class="refDate">22 September 2021</time>, <span>&lt;<a href="https://datatracker.ietf.org/doc/html/draft-ietf-nvo3-vxlan-gpe-12">https://datatracker.ietf.org/doc/html/draft-ietf-nvo3-vxlan-gpe-12</a>&gt;</span>. </dd>
3327 <dd class="break"></dd>
3328 </dl>
3329 </section>
3330 </section>
3331 <div id="sect-10">
3332 <section id="appendix-A">
3333 <h2 id="name-acknowledgements">
3334 <a href="#name-acknowledgements" class="section-name selfRef">Acknowledgements</a>
3335 </h2>
3336 <p id="appendix-A-1">
3337 The authors would like to thank <span class="contact-name">Sami Boutros</span>, <span class="contact-name">Jeffrey Zhang</span>,
3338 <span class="contact-name">Krzysztof Szarkowicz</span>, <span class="contact-name">Lukas Krattiger</span> and <span class="contact-name">Neeraj Malhotra</span> for their
3339 valuable comments. The authors would also like to thank <span class="contact-name">Linda Dunbar</span>, <span class="contact-name">Florin Balus</span>, <span class="contact-name">Yakov Rekhter</span>, <span class="contact-name">Wim Henderickx</span>, <span class="contact-name">Lucy Yong</span>, and
3340 <span class="contact-name">Dennis Cai</span> for their feedback and contributions.<a href="#appendix-A-1" class="pilcrow">¶</a></p>
3341 </section>
3342 </div>
3343 <div id="authors-addresses">
3344 <section id="appendix-B">
3345 <h2 id="name-authors-addresses">
3346 <a href="#name-authors-addresses" class="section-name selfRef">Authors' Addresses</a>
3347 </h2>
3348 <address class="vcard">
3349 <div dir="auto" class="left"><span class="fn nameRole">Ali Sajassi</span></div>
3350 <div dir="auto" class="left"><span class="org">Cisco Systems</span></div>
3351 <div class="email">
3352 <span>Email:</span>
3353 <a href="mailto:[email protected]" class="email">[email protected]</a>
3354 </div>
3355 </address>
3356 <address class="vcard">
3357 <div dir="auto" class="left"><span class="fn nameRole">Samer Salam</span></div>
3358 <div dir="auto" class="left"><span class="org">Cisco Systems</span></div>
3359 <div class="email">
3360 <span>Email:</span>
3361 <a href="mailto:[email protected]" class="email">[email protected]</a>
3362 </div>
3363 </address>
3364 <address class="vcard">
3365 <div dir="auto" class="left"><span class="fn nameRole">Samir Thoria</span></div>
3366 <div dir="auto" class="left"><span class="org">Cisco Systems</span></div>
3367 <div class="email">
3368 <span>Email:</span>
3369 <a href="mailto:[email protected]" class="email">[email protected]</a>
3370 </div>
3371 </address>
3372 <address class="vcard">
3373 <div dir="auto" class="left"><span class="fn nameRole">John E Drake</span></div>
3374 <div dir="auto" class="left"><span class="org">Juniper</span></div>
3375 <div class="email">
3376 <span>Email:</span>
3377 <a href="mailto:[email protected]" class="email">[email protected]</a>
3378 </div>
3379 </address>
3380 <address class="vcard">
3381 <div dir="auto" class="left"><span class="fn nameRole">Jorge Rabadan</span></div>
3382 <div dir="auto" class="left"><span class="org">Nokia</span></div>
3383 <div class="email">
3384 <span>Email:</span>
3385 <a href="mailto:[email protected]" class="email">[email protected]</a>
3386 </div>
3387 </address>
3388 </section>
3389 </div>
3390 <script>const toc = document.getElementById("toc");
3391 toc.querySelector("h2").addEventListener("click", e => {
3392 toc.classList.toggle("active");
3393 });
3394 toc.querySelector("nav").addEventListener("click", e => {
3395 toc.classList.remove("active");
3396 });
3397 </script>
3398 </body>
3399 </html>