Debian Codegraph @

rfc7365.html

raw
1 <pre>Internet Engineering Task Force (IETF) M. Lasserre
2 Request for Comments: 7365 F. Balus
3 Category: Informational Alcatel-Lucent
4 ISSN: 2070-1721 T. Morin
5 Orange
6 N. Bitar
7 Verizon
8 Y. Rekhter
9 Juniper
10 October 2014
11
12
13 <span class="h1">Framework for Data Center (DC) Network Virtualization</span>
14
15 Abstract
16
17 This document provides a framework for Data Center (DC) Network
18 Virtualization over Layer 3 (NVO3) and defines a reference model
19 along with logical components required to design a solution.
20
21 Status of This Memo
22
23 This document is not an Internet Standards Track specification; it is
24 published for informational purposes.
25
26 This document is a product of the Internet Engineering Task Force
27 (IETF). It represents the consensus of the IETF community. It has
28 received public review and has been approved for publication by the
29 Internet Engineering Steering Group (IESG). Not all documents
30 approved by the IESG are a candidate for any level of Internet
31 Standard; see <a href="./rfc5741#section-2">Section&nbsp;2 of RFC 5741</a>.
32
33 Information about the current status of this document, any errata,
34 and how to provide feedback on it may be obtained at
35 <a href="http://www.rfc-editor.org/info/rfc7365">http://www.rfc-editor.org/info/rfc7365</a>.
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52 <span class="grey">Lasserre, et al. Informational [Page 1]</span></pre>
53 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span>
54 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
55
56
57 Copyright Notice
58
59 Copyright (c) 2014 IETF Trust and the persons identified as the
60 document authors. All rights reserved.
61
62 This document is subject to <a href="https://www.rfc-editor.org/bcp/bcp78">BCP 78</a> and the IETF Trust's Legal
63 Provisions Relating to IETF Documents
64 (<a href="http://trustee.ietf.org/license-info">http://trustee.ietf.org/license-info</a>) in effect on the date of
65 publication of this document. Please review these documents
66 carefully, as they describe your rights and restrictions with respect
67 to this document. Code Components extracted from this document must
68 include Simplified BSD License text as described in Section 4.e of
69 the Trust Legal Provisions and are provided without warranty as
70 described in the Simplified BSD License.
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108 <span class="grey">Lasserre, et al. Informational [Page 2]</span></pre>
109 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span>
110 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
111
112
113 Table of Contents
114
115 <a href="#section-1">1</a>. Introduction ....................................................<a href="#page-4">4</a>
116 <a href="#section-1.1">1.1</a>. General Terminology ........................................<a href="#page-4">4</a>
117 <a href="#section-1.2">1.2</a>. DC Network Architecture ....................................<a href="#page-7">7</a>
118 <a href="#section-2">2</a>. Reference Models ................................................<a href="#page-8">8</a>
119 <a href="#section-2.1">2.1</a>. Generic Reference Model ....................................<a href="#page-8">8</a>
120 <a href="#section-2.2">2.2</a>. NVE Reference Model .......................................<a href="#page-10">10</a>
121 <a href="#section-2.3">2.3</a>. NVE Service Types .........................................<a href="#page-11">11</a>
122 <a href="#section-2.3.1">2.3.1</a>. L2 NVE Providing Ethernet LAN-Like Service .........<a href="#page-11">11</a>
123 <a href="#section-2.3.2">2.3.2</a>. L3 NVE Providing IP/VRF-Like Service ...............<a href="#page-11">11</a>
124 <a href="#section-2.4">2.4</a>. Operational Management Considerations .....................<a href="#page-12">12</a>
125 <a href="#section-3">3</a>. Functional Components ..........................................<a href="#page-12">12</a>
126 <a href="#section-3.1">3.1</a>. Service Virtualization Components .........................<a href="#page-12">12</a>
127 <a href="#section-3.1.1">3.1.1</a>. Virtual Access Points (VAPs) .......................<a href="#page-12">12</a>
128 <a href="#section-3.1.2">3.1.2</a>. Virtual Network Instance (VNI) .....................<a href="#page-12">12</a>
129 <a href="#section-3.1.3">3.1.3</a>. Overlay Modules and VN Context .....................<a href="#page-14">14</a>
130 <a href="#section-3.1.4">3.1.4</a>. Tunnel Overlays and Encapsulation Options ..........<a href="#page-14">14</a>
131 <a href="#section-3.1.5">3.1.5</a>. Control-Plane Components ...........................<a href="#page-14">14</a>
132 3.1.5.1. Distributed vs. Centralized
133 Control Plane .............................<a href="#page-14">14</a>
134 <a href="#section-3.1.5.2">3.1.5.2</a>. Auto-provisioning and Service Discovery ...<a href="#page-15">15</a>
135 <a href="#section-3.1.5.3">3.1.5.3</a>. Address Advertisement and Tunnel Mapping ..15
136 <a href="#section-3.1.5.4">3.1.5.4</a>. Overlay Tunneling .........................<a href="#page-16">16</a>
137 <a href="#section-3.2">3.2</a>. Multihoming ...............................................<a href="#page-16">16</a>
138 <a href="#section-3.3">3.3</a>. VM Mobility ...............................................<a href="#page-17">17</a>
139 <a href="#section-4">4</a>. Key Aspects of Overlay Networks ................................<a href="#page-17">17</a>
140 <a href="#section-4.1">4.1</a>. Pros and Cons .............................................<a href="#page-18">18</a>
141 <a href="#section-4.2">4.2</a>. Overlay Issues to Consider ................................<a href="#page-19">19</a>
142 <a href="#section-4.2.1">4.2.1</a>. Data Plane vs. Control Plane Driven ................<a href="#page-19">19</a>
143 <a href="#section-4.2.2">4.2.2</a>. Coordination between Data Plane and Control Plane ..19
144 4.2.3. Handling Broadcast, Unknown Unicast, and
145 Multicast (BUM) Traffic ............................<a href="#page-20">20</a>
146 <a href="#section-4.2.4">4.2.4</a>. Path MTU ...........................................<a href="#page-20">20</a>
147 <a href="#section-4.2.5">4.2.5</a>. NVE Location Trade-Offs ............................<a href="#page-21">21</a>
148 4.2.6. Interaction between Network Overlays and
149 Underlays ..........................................<a href="#page-22">22</a>
150 <a href="#section-5">5</a>. Security Considerations ........................................<a href="#page-22">22</a>
151 <a href="#section-6">6</a>. Informative References .........................................<a href="#page-24">24</a>
152 Acknowledgments ...................................................<a href="#page-26">26</a>
153 Authors' Addresses ................................................<a href="#page-26">26</a>
154
155
156
157
158
159
160
161
162
163
164 <span class="grey">Lasserre, et al. Informational [Page 3]</span></pre>
165 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span>
166 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
167
168
169 <span class="h2"><a class="selflink" id="section-1" href="#section-1">1</a>. Introduction</span>
170
171 This document provides a framework for Data Center (DC) Network
172 Virtualization over Layer 3 (NVO3) tunnels. This framework is
173 intended to aid in standardizing protocols and mechanisms to support
174 large-scale network virtualization for data centers.
175
176 [<a id="ref-RFC7364">RFC7364</a>] defines the rationale for using overlay networks in order
177 to build large multi-tenant data center networks. Compute, storage
178 and network virtualization are often used in these large data centers
179 to support a large number of communication domains and end systems.
180
181 This document provides reference models and functional components of
182 data center overlay networks as well as a discussion of technical
183 issues that have to be addressed.
184
185 <span class="h3"><a class="selflink" id="section-1.1" href="#section-1.1">1.1</a>. General Terminology</span>
186
187 This document uses the following terminology:
188
189 NVO3 Network: An overlay network that provides a Layer 2 (L2) or
190 Layer 3 (L3) service to Tenant Systems over an L3 underlay network
191 using the architecture and protocols as defined by the NVO3 Working
192 Group.
193
194 Network Virtualization Edge (NVE): An NVE is the network entity that
195 sits at the edge of an underlay network and implements L2 and/or L3
196 network virtualization functions. The network-facing side of the NVE
197 uses the underlying L3 network to tunnel tenant frames to and from
198 other NVEs. The tenant-facing side of the NVE sends and receives
199 Ethernet frames to and from individual Tenant Systems. An NVE could
200 be implemented as part of a virtual switch within a hypervisor, a
201 physical switch or router, or a Network Service Appliance, or it
202 could be split across multiple devices.
203
204 Virtual Network (VN): A VN is a logical abstraction of a physical
205 network that provides L2 or L3 network services to a set of Tenant
206 Systems. A VN is also known as a Closed User Group (CUG).
207
208 Virtual Network Instance (VNI): A specific instance of a VN from the
209 perspective of an NVE.
210
211 Virtual Network Context (VN Context) Identifier: Field in an overlay
212 encapsulation header that identifies the specific VN the packet
213 belongs to. The egress NVE uses the VN Context identifier to deliver
214 the packet to the correct Tenant System. The VN Context identifier
215 can be a locally significant identifier or a globally unique
216 identifier.
217
218
219
220 <span class="grey">Lasserre, et al. Informational [Page 4]</span></pre>
221 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span>
222 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
223
224
225 Underlay or Underlying Network: The network that provides the
226 connectivity among NVEs and that NVO3 packets are tunneled over,
227 where an NVO3 packet carries an NVO3 overlay header followed by a
228 tenant packet. The underlay network does not need to be aware that
229 it is carrying NVO3 packets. Addresses on the underlay network
230 appear as "outer addresses" in encapsulated NVO3 packets. In
231 general, the underlay network can use a completely different protocol
232 (and address family) from that of the overlay. In the case of NVO3,
233 the underlay network is IP.
234
235 Data Center (DC): A physical complex housing physical servers,
236 network switches and routers, network service appliances, and
237 networked storage. The purpose of a data center is to provide
238 application, compute, and/or storage services. One such service is
239 virtualized infrastructure data center services, also known as
240 "Infrastructure as a Service".
241
242 Virtual Data Center (Virtual DC): A container for virtualized
243 compute, storage, and network services. A virtual DC is associated
244 with a single tenant and can contain multiple VNs and Tenant Systems
245 connected to one or more of these VNs.
246
247 Virtual Machine (VM): A software implementation of a physical machine
248 that runs programs as if they were executing on a physical, non-
249 virtualized machine. Applications (generally) do not know they are
250 running on a VM as opposed to running on a "bare metal" host or
251 server, though some systems provide a para-virtualization environment
252 that allows an operating system or application to be aware of the
253 presence of virtualization for optimization purposes.
254
255 Hypervisor: Software running on a server that allows multiple VMs to
256 run on the same physical server. The hypervisor manages and provides
257 shared computation, memory, and storage services and network
258 connectivity to the VMs that it hosts. Hypervisors often embed a
259 virtual switch (see below).
260
261 Server: A physical end-host machine that runs user applications. A
262 standalone (or "bare metal") server runs a conventional operating
263 system hosting a single-tenant application. A virtualized server
264 runs a hypervisor supporting one or more VMs.
265
266 Virtual Switch (vSwitch): A function within a hypervisor (typically
267 implemented in software) that provides similar forwarding services to
268 a physical Ethernet switch. A vSwitch forwards Ethernet frames
269 between VMs running on the same server or between a VM and a physical
270 Network Interface Card (NIC) connecting the server to a physical
271 Ethernet switch or router. A vSwitch also enforces network isolation
272 between VMs that by policy are not permitted to communicate with each
273
274
275
276 <span class="grey">Lasserre, et al. Informational [Page 5]</span></pre>
277 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-6" ></span>
278 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
279
280
281 other (e.g., by honoring VLANs). A vSwitch may be bypassed when an
282 NVE is enabled on the host server.
283
284 Tenant: The customer using a virtual network and any associated
285 resources (e.g., compute, storage, and network). A tenant could be
286 an enterprise or a department/organization within an enterprise.
287
288 Tenant System: A physical or virtual system that can play the role of
289 a host or a forwarding element such as a router, switch, firewall,
290 etc. It belongs to a single tenant and connects to one or more VNs
291 of that tenant.
292
293 Tenant Separation: Refers to isolating traffic of different tenants
294 such that traffic from one tenant is not visible to or delivered to
295 another tenant, except when allowed by policy. Tenant separation
296 also refers to address space separation, whereby different tenants
297 can use the same address space without conflict.
298
299 Virtual Access Points (VAPs): A logical connection point on the NVE
300 for connecting a Tenant System to a virtual network. Tenant Systems
301 connect to VNIs at an NVE through VAPs. VAPs can be physical ports
302 or virtual ports identified through logical interface identifiers
303 (e.g., VLAN ID or internal vSwitch Interface ID connected to a VM).
304
305 End Device: A physical device that connects directly to the DC
306 underlay network. This is in contrast to a Tenant System, which
307 connects to a corresponding tenant VN. An End Device is administered
308 by the DC operator rather than a tenant and is part of the DC
309 infrastructure. An End Device may implement NVO3 technology in
310 support of NVO3 functions. Examples of an End Device include hosts
311 (e.g., server or server blade), storage systems (e.g., file servers
312 and iSCSI storage systems), and network devices (e.g., firewall,
313 load-balancer, and IPsec gateway).
314
315 Network Virtualization Authority (NVA): Entity that provides
316 reachability and forwarding information to NVEs.
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332 <span class="grey">Lasserre, et al. Informational [Page 6]</span></pre>
333 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-7" ></span>
334 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
335
336
337 <span class="h3"><a class="selflink" id="section-1.2" href="#section-1.2">1.2</a>. DC Network Architecture</span>
338
339 A generic architecture for data centers is depicted in Figure 1:
340
341 ,---------.
342 ,' `.
343 ( IP/MPLS WAN )
344 `. ,'
345 `-+------+'
346 \ /
347 +--------+ +--------+
348 | DC |+-+| DC |
349 |gateway |+-+|gateway |
350 +--------+ +--------+
351 | /
352 .--. .--.
353 ( ' '.--.
354 .-.' Intra-DC '
355 ( network )
356 ( .'-'
357 '--'._.'. )\ \
358 / / '--' \ \
359 / / | | \ \
360 +--------+ +--------+ +--------+
361 | access | | access | | access |
362 | switch | | switch | | switch |
363 +--------+ +--------+ +--------+
364 / \ / \ / \
365 __/_ \ / \ /_ _\__
366 '--------' '--------' '--------' '--------'
367 : End : : End : : End : : End :
368 : Device : : Device : : Device : : Device :
369 '--------' '--------' '--------' '--------'
370
371 Figure 1: A Generic Architecture for Data Centers
372
373 An example of multi-tier DC network architecture is presented in
374 Figure 1. It provides a view of the physical components inside a DC.
375
376 A DC network is usually composed of intra-DC networks and network
377 services, and inter-DC network and network connectivity services.
378
379 DC networking elements can act as strict L2 switches and/or provide
380 IP routing capabilities, including network service virtualization.
381
382 In some DC architectures, some tier layers could provide L2 and/or L3
383 services. In addition, some tier layers may be collapsed, and
384 Internet connectivity, inter-DC connectivity, and VPN support may be
385
386
387
388 <span class="grey">Lasserre, et al. Informational [Page 7]</span></pre>
389 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-8" ></span>
390 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
391
392
393 handled by a smaller number of nodes. Nevertheless, one can assume
394 that the network functional blocks in a DC fit in the architecture
395 depicted in Figure 1.
396
397 The following components can be present in a DC:
398
399 - Access switch: Hardware-based Ethernet switch aggregating all
400 Ethernet links from the End Devices in a rack representing the
401 entry point in the physical DC network for the hosts. It may also
402 provide routing functionality, virtual IP network connectivity, or
403 Layer 2 tunneling over IP, for instance. Access switches are
404 usually multihomed to aggregation switches in the Intra-DC
405 network. A typical example of an access switch is a Top-of-Rack
406 (ToR) switch. Other deployment scenarios may use an intermediate
407 Blade Switch before the ToR, or an End-of-Row (EoR) switch, to
408 provide similar functions to a ToR.
409
410 - Intra-DC Network: Network composed of high-capacity core nodes
411 (Ethernet switches/routers). Core nodes may provide virtual
412 Ethernet bridging and/or IP routing services.
413
414 - DC Gateway (DC GW): Gateway to the outside world providing DC
415 interconnect and connectivity to Internet and VPN customers. In
416 the current DC network model, this may be simply a router
417 connected to the Internet and/or an IP VPN/L2VPN PE. Some network
418 implementations may dedicate DC GWs for different connectivity
419 types (e.g., a DC GW for Internet and another for VPN).
420
421 Note that End Devices may be single-homed or multihomed to access
422 switches.
423
424 <span class="h2"><a class="selflink" id="section-2" href="#section-2">2</a>. Reference Models</span>
425
426 <span class="h3"><a class="selflink" id="section-2.1" href="#section-2.1">2.1</a>. Generic Reference Model</span>
427
428 Figure 2 depicts a DC reference model for network virtualization
429 overlays where NVEs provide a logical interconnect between Tenant
430 Systems that belong to a specific VN.
431
432
433
434
435
436
437
438
439
440
441
442
443
444 <span class="grey">Lasserre, et al. Informational [Page 8]</span></pre>
445 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-9" ></span>
446 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
447
448
449 +--------+ +--------+
450 | Tenant +--+ +----| Tenant |
451 | System | | (') | System |
452 +--------+ | ................. ( ) +--------+
453 | +---+ +---+ (_)
454 +--|NVE|---+ +---|NVE|-----+
455 +---+ | | +---+
456 / . +-----+ .
457 / . +--| NVA |--+ .
458 / . | +-----+ \ .
459 | . | \ .
460 | . | Overlay +--+--++--------+
461 +--------+ | . | Network | NVE || Tenant |
462 | Tenant +--+ . | | || System |
463 | System | . \ +---+ +--+--++--------+
464 +--------+ .....|NVE|.........
465 +---+
466 |
467 |
468 =====================
469 | |
470 +--------+ +--------+
471 | Tenant | | Tenant |
472 | System | | System |
473 +--------+ +--------+
474
475 Figure 2: Generic Reference Model for DC Network Virtualization
476 Overlays
477
478 In order to obtain reachability information, NVEs may exchange
479 information directly between themselves via a control-plane protocol.
480 In this case, a control-plane module resides in every NVE.
481
482 It is also possible for NVEs to communicate with an external Network
483 Virtualization Authority (NVA) to obtain reachability and forwarding
484 information. In this case, a protocol is used between NVEs and
485 NVA(s) to exchange information.
486
487 It should be noted that NVAs may be organized in clusters for
488 redundancy and scalability and can appear as one logically
489 centralized controller. In this case, inter-NVA communication is
490 necessary to synchronize state among nodes within a cluster or share
491 information across clusters. The information exchanged between NVAs
492 of the same cluster could be different from the information exchanged
493 across clusters.
494
495
496
497
498
499
500 <span class="grey">Lasserre, et al. Informational [Page 9]</span></pre>
501 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-10" ></span>
502 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
503
504
505 A Tenant System can be attached to an NVE in several ways:
506
507 - locally, by being co-located in the same End Device
508
509 - remotely, via a point-to-point connection or a switched network
510
511 When an NVE is co-located with a Tenant System, the state of the
512 Tenant System can be determined without protocol assistance. For
513 instance, the operational status of a VM can be communicated via a
514 local API. When an NVE is remotely connected to a Tenant System, the
515 state of the Tenant System or NVE needs to be exchanged directly or
516 via a management entity, using a control-plane protocol or API, or
517 directly via a data-plane protocol.
518
519 The functional components in Figure 2 do not necessarily map directly
520 to the physical components described in Figure 1. For example, an
521 End Device can be a server blade with VMs and a virtual switch. A VM
522 can be a Tenant System, and the NVE functions may be performed by the
523 host server. In this case, the Tenant System and NVE function are
524 co-located. Another example is the case where the End Device is the
525 Tenant System and the NVE function can be implemented by the
526 connected ToR. In this case, the Tenant System and NVE function are
527 not co-located.
528
529 Underlay nodes utilize L3 technologies to interconnect NVE nodes.
530 These nodes perform forwarding based on outer L3 header information,
531 and generally do not maintain state for each tenant service, albeit
532 some applications (e.g., multicast) may require control-plane or
533 forwarding-plane information that pertains to a tenant, group of
534 tenants, tenant service, or a set of services that belong to one or
535 more tenants. Mechanisms to control the amount of state maintained
536 in the underlay may be needed.
537
538 <span class="h3"><a class="selflink" id="section-2.2" href="#section-2.2">2.2</a>. NVE Reference Model</span>
539
540 Figure 3 depicts the NVE reference model. One or more VNIs can be
541 instantiated on an NVE. A Tenant System interfaces with a
542 corresponding VNI via a VAP. An overlay module provides tunneling
543 overlay functions (e.g., encapsulation and decapsulation of tenant
544 traffic, tenant identification, and mapping, etc.).
545
546
547
548
549
550
551
552
553
554
555
556 <span class="grey">Lasserre, et al. Informational [Page 10]</span></pre>
557 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-11" ></span>
558 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
559
560
561 +-------- L3 Network -------+
562 | |
563 | Tunnel Overlay |
564 +------------+---------+ +---------+------------+
565 | +----------+-------+ | | +---------+--------+ |
566 | | Overlay Module | | | | Overlay Module | |
567 | +---------+--------+ | | +---------+--------+ |
568 | |VN Context| | VN Context| |
569 | | | | | |
570 | +--------+-------+ | | +--------+-------+ |
571 | | |VNI| . |VNI| | | | |VNI| . |VNI| |
572 NVE1 | +-+------------+-+ | | +-+-----------+--+ | NVE2
573 | | VAPs | | | | VAPs | |
574 +----+------------+----+ +----+-----------+-----+
575 | | | |
576 | | | |
577 Tenant Systems Tenant Systems
578
579 Figure 3: Generic NVE Reference Model
580
581 Note that some NVE functions (e.g., data-plane and control-plane
582 functions) may reside in one device or may be implemented separately
583 in different devices.
584
585 <span class="h3"><a class="selflink" id="section-2.3" href="#section-2.3">2.3</a>. NVE Service Types</span>
586
587 An NVE provides different types of virtualized network services to
588 multiple tenants, i.e., an L2 service or an L3 service. Note that an
589 NVE may be capable of providing both L2 and L3 services for a tenant.
590 This section defines the service types and associated attributes.
591
592 <span class="h4"><a class="selflink" id="section-2.3.1" href="#section-2.3.1">2.3.1</a>. L2 NVE Providing Ethernet LAN-Like Service</span>
593
594 An L2 NVE implements Ethernet LAN emulation, an Ethernet-based
595 multipoint service similar to an IETF Virtual Private LAN Service
596 (VPLS) [<a href="./rfc4761" title="&quot;Virtual Private LAN Service (VPLS) Using BGP for Auto-Discovery and Signaling&quot;">RFC4761</a>][RFC4762] or Ethernet VPN [<a href="#ref-EVPN" title="&quot;BGP MPLS Based Ethernet VPN&quot;">EVPN</a>] service, where the
597 Tenant Systems appear to be interconnected by a LAN environment over
598 an L3 overlay. As such, an L2 NVE provides per-tenant virtual
599 switching instance (L2 VNI) and L3 (IP/MPLS) tunneling encapsulation
600 of tenant Media Access Control (MAC) frames across the underlay.
601 Note that the control plane for an L2 NVE could be implemented
602 locally on the NVE or in a separate control entity.
603
604 <span class="h4"><a class="selflink" id="section-2.3.2" href="#section-2.3.2">2.3.2</a>. L3 NVE Providing IP/VRF-Like Service</span>
605
606 An L3 NVE provides virtualized IP forwarding service, similar to IETF
607 IP VPN (e.g., BGP/MPLS IP VPN [<a href="./rfc4364" title="&quot;BGP/MPLS IP Virtual Private Networks (VPNs)&quot;">RFC4364</a>]) from a service definition
608 perspective. That is, an L3 NVE provides per-tenant forwarding and
609
610
611
612 <span class="grey">Lasserre, et al. Informational [Page 11]</span></pre>
613 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-12" ></span>
614 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
615
616
617 routing instance (L3 VNI) and L3 (IP/MPLS) tunneling encapsulation of
618 tenant IP packets across the underlay. Note that routing could be
619 performed locally on the NVE or in a separate control entity.
620
621 <span class="h3"><a class="selflink" id="section-2.4" href="#section-2.4">2.4</a>. Operational Management Considerations</span>
622
623 NVO3 services are overlay services over an IP underlay.
624
625 As far as the IP underlay is concerned, existing IP Operations,
626 Administration, and Maintenance (OAM) facilities are used.
627
628 With regard to the NVO3 overlay, both L2 and L3 services can be
629 offered. It is expected that existing fault and performance OAM
630 facilities will be used. Sections <a href="#section-4.1">4.1</a> and <a href="#section-4.2.6">4.2.6</a> provide further
631 discussion of additional fault and performance management issues to
632 consider.
633
634 As far as configuration is concerned, the DC environment is driven by
635 the need to bring new services up rapidly and is typically very
636 dynamic, specifically in the context of virtualized services. It is
637 therefore critical to automate the configuration of NVO3 services.
638
639 <span class="h2"><a class="selflink" id="section-3" href="#section-3">3</a>. Functional Components</span>
640
641 This section decomposes the network virtualization architecture into
642 the functional components described in Figure 3 to make it easier to
643 discuss solution options for these components.
644
645 <span class="h3"><a class="selflink" id="section-3.1" href="#section-3.1">3.1</a>. Service Virtualization Components</span>
646
647 <span class="h4"><a class="selflink" id="section-3.1.1" href="#section-3.1.1">3.1.1</a>. Virtual Access Points (VAPs)</span>
648
649 Tenant Systems are connected to VNIs through Virtual Access Points
650 (VAPs).
651
652 VAPs can be physical ports or virtual ports identified through
653 logical interface identifiers (e.g., VLAN ID and internal vSwitch
654 Interface ID connected to a VM).
655
656 <span class="h4"><a class="selflink" id="section-3.1.2" href="#section-3.1.2">3.1.2</a>. Virtual Network Instance (VNI)</span>
657
658 A VNI is a specific VN instance on an NVE. Each VNI defines a
659 forwarding context that contains reachability information and
660 policies.
661
662
663
664
665
666
667
668 <span class="grey">Lasserre, et al. Informational [Page 12]</span></pre>
669 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-13" ></span>
670 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
671
672
673 <span class="h4"><a class="selflink" id="section-3.1.3" href="#section-3.1.3">3.1.3</a>. Overlay Modules and VN Context</span>
674
675 Mechanisms for identifying each tenant service are required to allow
676 the simultaneous overlay of multiple tenant services over the same
677 underlay L3 network topology. In the data plane, each NVE, upon
678 sending a tenant packet, must be able to encode the VN Context for
679 the destination NVE in addition to the L3 tunneling information
680 (e.g., source IP address identifying the source NVE and the
681 destination IP address identifying the destination NVE, or MPLS
682 label). This allows the destination NVE to identify the tenant
683 service instance and therefore appropriately process and forward the
684 tenant packet.
685
686 The overlay module provides tunneling overlay functions: tunnel
687 initiation/termination as in the case of stateful tunnels (see
688 <a href="#section-3.1.4">Section 3.1.4</a>) and/or encapsulation/decapsulation of frames from the
689 VAPs/L3 underlay.
690
691 In a multi-tenant context, tunneling aggregates frames from/to
692 different VNIs. Tenant identification and traffic demultiplexing are
693 based on the VN Context identifier.
694
695 The following approaches can be considered:
696
697 - VN Context identifier per Tenant: This is a globally unique (on a
698 per-DC administrative domain) VN identifier used to identify the
699 corresponding VNI. Examples of such identifiers in existing
700 technologies are IEEE VLAN IDs and Service Instance IDs (I-SIDs)
701 that identify virtual L2 domains when using IEEE 802.1Q and IEEE
702 802.1ah, respectively. Note that multiple VN identifiers can
703 belong to a tenant.
704
705 - One VN Context identifier per VNI: Each VNI value is automatically
706 generated by the egress NVE, or a control plane associated with
707 that NVE, and usually distributed by a control-plane protocol to
708 all the related NVEs. An example of this approach is the use of
709 per-VRF MPLS labels in IP VPN [<a href="./rfc4364" title="&quot;BGP/MPLS IP Virtual Private Networks (VPNs)&quot;">RFC4364</a>]. The VNI value is
710 therefore locally significant to the egress NVE.
711
712 - One VN Context identifier per VAP: A value locally significant to
713 an NVE is assigned and usually distributed by a control-plane
714 protocol to identify a VAP. An example of this approach is the
715 use of per-CE MPLS labels in IP VPN [<a href="./rfc4364" title="&quot;BGP/MPLS IP Virtual Private Networks (VPNs)&quot;">RFC4364</a>].
716
717 Note that when using one VN Context per VNI or per VAP, an additional
718 global identifier (e.g., a VN identifier or name) may be used by the
719 control plane to identify the tenant context.
720
721
722
723
724 <span class="grey">Lasserre, et al. Informational [Page 13]</span></pre>
725 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-14" ></span>
726 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
727
728
729 <span class="h4"><a class="selflink" id="section-3.1.4" href="#section-3.1.4">3.1.4</a>. Tunnel Overlays and Encapsulation Options</span>
730
731 Once the VN Context identifier is added to the frame, an L3 tunnel
732 encapsulation is used to transport the frame to the destination NVE.
733
734 Different IP tunneling options (e.g., Generic Routing Encapsulation
735 (GRE), the Layer 2 Tunneling Protocol (L2TP), and IPsec) and MPLS
736 tunneling can be used. Tunneling could be stateless or stateful.
737 Stateless tunneling simply entails the encapsulation of a tenant
738 packet with another header necessary for forwarding the packet across
739 the underlay (e.g., IP tunneling over an IP underlay). Stateful
740 tunneling, on the other hand, entails maintaining tunneling state at
741 the tunnel endpoints (i.e., NVEs). Tenant packets on an ingress NVE
742 can then be transmitted over such tunnels to a destination (egress)
743 NVE by encapsulating the packets with a corresponding tunneling
744 header. The tunneling state at the endpoints may be configured or
745 dynamically established. Solutions should specify the tunneling
746 technology used and whether it is stateful or stateless. In this
747 document, however, tunneling and tunneling encapsulation are used
748 interchangeably to simply mean the encapsulation of a tenant packet
749 with a tunneling header necessary to carry the packet between an
750 ingress NVE and an egress NVE across the underlay. It should be
751 noted that stateful tunneling, especially when configuration is
752 involved, does impose management overhead and scale constraints.
753 When confidentiality is required, the use of opportunistic security
754 [<a href="#ref-OPPSEC" title="&quot;Opportunistic Security: Some Protection Most of the Time&quot;">OPPSEC</a>] can be used as a stateless tunneling solution.
755
756 <span class="h4"><a class="selflink" id="section-3.1.5" href="#section-3.1.5">3.1.5</a>. Control-Plane Components</span>
757
758 <span class="h5"><a class="selflink" id="section-3.1.5.1" href="#section-3.1.5.1">3.1.5.1</a>. Distributed vs. Centralized Control Plane</span>
759
760 Control- and management-plane entities can be centralized or
761 distributed. Both approaches have been used extensively in the past.
762 The routing model of the Internet is a good example of a distributed
763 approach. Transport networks have usually used a centralized
764 approach to manage transport paths.
765
766 It is also possible to combine the two approaches, i.e., using a
767 hybrid model. A global view of network state can have many benefits,
768 but it does not preclude the use of distributed protocols within the
769 network. Centralized models provide a facility to maintain global
770 state and distribute that state to the network. When used in
771 combination with distributed protocols, greater network efficiencies,
772 improved reliability, and robustness can be achieved. Domain- and/or
773 deployment-specific constraints define the balance between
774 centralized and distributed approaches.
775
776
777
778
779
780 <span class="grey">Lasserre, et al. Informational [Page 14]</span></pre>
781 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-15" ></span>
782 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
783
784
785 <span class="h5"><a class="selflink" id="section-3.1.5.2" href="#section-3.1.5.2">3.1.5.2</a>. Auto-provisioning and Service Discovery</span>
786
787 NVEs must be able to identify the appropriate VNI for each Tenant
788 System. This is based on state information that is often provided by
789 external entities. For example, in an environment where a VM is a
790 Tenant System, this information is provided by VM orchestration
791 systems, since these are the only entities that have visibility of
792 which VM belongs to which tenant.
793
794 A mechanism for communicating this information to the NVE is
795 required. VAPs have to be created and mapped to the appropriate VNI.
796 Depending upon the implementation, this control interface can be
797 implemented using an auto-discovery protocol between Tenant Systems
798 and their local NVE or through management entities. In either case,
799 appropriate security and authentication mechanisms to verify that
800 Tenant System information is not spoofed or altered are required.
801 This is one critical aspect for providing integrity and tenant
802 isolation in the system.
803
804 NVEs may learn reachability information for VNIs on other NVEs via a
805 control protocol that exchanges such information among NVEs or via a
806 management-control entity.
807
808 <span class="h5"><a class="selflink" id="section-3.1.5.3" href="#section-3.1.5.3">3.1.5.3</a>. Address Advertisement and Tunnel Mapping</span>
809
810 As traffic reaches an ingress NVE on a VAP, a lookup is performed to
811 determine which NVE or local VAP the packet needs to be sent to. If
812 the packet is to be sent to another NVE, the packet is encapsulated
813 with a tunnel header containing the destination information
814 (destination IP address or MPLS label) of the egress NVE.
815 Intermediate nodes (between the ingress and egress NVEs) switch or
816 route traffic based upon the tunnel destination information.
817
818 A key step in the above process consists of identifying the
819 destination NVE the packet is to be tunneled to. NVEs are
820 responsible for maintaining a set of forwarding or mapping tables
821 that hold the bindings between destination VM and egress NVE
822 addresses. Several ways of populating these tables are possible:
823 control plane driven, management plane driven, or data plane driven.
824
825 When a control-plane protocol is used to distribute address
826 reachability and tunneling information, the auto-provisioning and
827 service discovery could be accomplished by the same protocol. In
828 this scenario, the auto-provisioning and service discovery could be
829 combined with (be inferred from) the address advertisement and
830 associated tunnel mapping. Furthermore, a control-plane protocol
831
832
833
834
835
836 <span class="grey">Lasserre, et al. Informational [Page 15]</span></pre>
837 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-16" ></span>
838 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
839
840
841 that carries both MAC and IP addresses eliminates the need for the
842 Address Resolution Protocol (ARP) and hence addresses one of the
843 issues with explosive ARP handling as discussed in [<a href="./rfc6820" title="&quot;Address Resolution Problems in Large Data Center Networks&quot;">RFC6820</a>].
844
845 <span class="h5"><a class="selflink" id="section-3.1.5.4" href="#section-3.1.5.4">3.1.5.4</a>. Overlay Tunneling</span>
846
847 For overlay tunneling, and dependent upon the tunneling technology
848 used for encapsulating the Tenant System packets, it may be
849 sufficient to have one or more local NVE addresses assigned and used
850 in the source and destination fields of a tunneling encapsulation
851 header. Other information that is part of the tunneling
852 encapsulation header may also need to be configured. In certain
853 cases, local NVE configuration may be sufficient while in other
854 cases, some tunneling-related information may need to be shared among
855 NVEs. The information that needs to be shared will be technology
856 dependent. For instance, potential information could include tunnel
857 identity, encapsulation type, and/or tunnel resources. In certain
858 cases, such as when using IP multicast in the underlay, tunnels that
859 interconnect NVEs may need to be established. When tunneling
860 information needs to be exchanged or shared among NVEs, a control-
861 plane protocol may be required. For instance, it may be necessary to
862 provide active/standby status information between NVEs, up/down
863 status information, pruning/grafting information for multicast
864 tunnels, etc.
865
866 In addition, a control plane may be required to set up the tunnel
867 path for some tunneling technologies. This applies to both unicast
868 and multicast tunneling.
869
870 <span class="h3"><a class="selflink" id="section-3.2" href="#section-3.2">3.2</a>. Multihoming</span>
871
872 Multihoming techniques can be used to increase the reliability of an
873 NVO3 network. It is also important to ensure that the physical
874 diversity in an NVO3 network is taken into account to avoid single
875 points of failure.
876
877 Multihoming can be enabled in various nodes, from Tenant Systems into
878 ToRs, ToRs into core switches/routers, and core nodes into DC GWs.
879
880 The NVO3 underlay nodes (i.e., from NVEs to DC GWs) rely on IP
881 routing techniques or MPLS re-rerouting capabilities as the means to
882 re-route traffic upon failures.
883
884 When a Tenant System is co-located with the NVE, the Tenant System is
885 effectively single-homed to the NVE via a virtual port. When the
886 Tenant System and the NVE are separated, the Tenant System is
887 connected to the NVE via a logical L2 construct such as a VLAN, and
888 it can be multihomed to various NVEs. An NVE may provide an L2
889
890
891
892 <span class="grey">Lasserre, et al. Informational [Page 16]</span></pre>
893 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-17" ></span>
894 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
895
896
897 service to the end system or an l3 service. An NVE may be multihomed
898 to a next layer in the DC at L2 or L3. When an NVE provides an L2
899 service and is not co-located with the end system, loop-avoidance
900 techniques must be used. Similarly, when the NVE provides L3
901 service, similar dual-homing techniques can be used. When the NVE
902 provides an L3 service to the end system, it is possible that no
903 dynamic routing protocol is enabled between the end system and the
904 NVE. The end system can be multihomed to multiple physically
905 separated L3 NVEs over multiple interfaces. When one of the links
906 connected to an NVE fails, the other interfaces can be used to reach
907 the end system.
908
909 External connectivity from a DC can be handled by two or more DC
910 gateways. Each gateway provides access to external networks such as
911 VPNs or the Internet. A gateway may be connected to two or more edge
912 nodes in the external network for redundancy. When a connection to
913 an upstream node is lost, the alternative connection is used, and the
914 failed route withdrawn.
915
916 <span class="h3"><a class="selflink" id="section-3.3" href="#section-3.3">3.3</a>. VM Mobility</span>
917
918 In DC environments utilizing VM technologies, an important feature is
919 that VMs can move from one server to another server in the same or
920 different L2 physical domains (within or across DCs) in a seamless
921 manner.
922
923 A VM can be moved from one server to another in stopped or suspended
924 state ("cold" VM mobility) or in running/active state ("hot" VM
925 mobility). With "hot" mobility, VM L2 and L3 addresses need to be
926 preserved. With "cold" mobility, it may be desired to preserve at
927 least VM L3 addresses.
928
929 Solutions to maintain connectivity while a VM is moved are necessary
930 in the case of "hot" mobility. This implies that connectivity among
931 VMs is preserved. For instance, for L2 VNs, ARP caches are updated
932 accordingly.
933
934 Upon VM mobility, NVE policies that define connectivity among VMs
935 must be maintained.
936
937 During VM mobility, it is expected that the path to the VM's default
938 gateway assures adequate QoS to VM applications, i.e., QoS that
939 matches the expected service-level agreement for these applications.
940
941 <span class="h2"><a class="selflink" id="section-4" href="#section-4">4</a>. Key Aspects of Overlay Networks</span>
942
943 The intent of this section is to highlight specific issues that
944 proposed overlay solutions need to address.
945
946
947
948 <span class="grey">Lasserre, et al. Informational [Page 17]</span></pre>
949 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-18" ></span>
950 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
951
952
953 <span class="h3"><a class="selflink" id="section-4.1" href="#section-4.1">4.1</a>. Pros and Cons</span>
954
955 An overlay network is a layer of virtual network topology on top of
956 the physical network.
957
958 Overlay networks offer the following key advantages:
959
960 - Unicast tunneling state management and association of Tenant
961 Systems reachability are handled at the edge of the network (at
962 the NVE). Intermediate transport nodes are unaware of such state.
963 Note that when multicast is enabled in the underlay network to
964 build multicast trees for tenant VNs, there would be more state
965 related to tenants in the underlay core network.
966
967 - Tunneling is used to aggregate traffic and hide tenant addresses
968 from the underlay network and hence offers the advantage of
969 minimizing the amount of forwarding state required within the
970 underlay network.
971
972 - Decoupling of the overlay addresses (MAC and IP) used by VMs from
973 the underlay network provides tenant separation and separation of
974 the tenant address spaces from the underlay address space.
975
976
977 - Overlay networks support of a large number of virtual network
978 identifiers.
979
980 Overlay networks also create several challenges:
981
982 - Overlay networks typically have no control of underlay networks
983 and lack underlay network information (e.g., underlay
984 utilization):
985
986 o Overlay networks and/or their associated management entities
987 typically probe the network to measure link or path properties,
988 such as available bandwidth or packet loss rate. It is
989 difficult to accurately evaluate network properties. It might
990 be preferable for the underlay network to expose usage and
991 performance information.
992
993 o Miscommunication or lack of coordination between overlay and
994 underlay networks can lead to an inefficient usage of network
995 resources.
996
997 o When multiple overlays co-exist on top of a common underlay
998 network, the lack of coordination between overlays can lead to
999 performance issues and/or resource usage inefficiencies.
1000
1001
1002
1003
1004 <span class="grey">Lasserre, et al. Informational [Page 18]</span></pre>
1005 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-19" ></span>
1006 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
1007
1008
1009 - Traffic carried over an overlay might fail to traverse firewalls
1010 and NAT devices.
1011
1012 - Multicast service scalability: Multicast support may be required
1013 in the underlay network to address tenant flood containment or
1014 efficient multicast handling. The underlay may also be required
1015 to maintain multicast state on a per-tenant basis or even on a
1016 per-individual multicast flow of a given tenant. Ingress
1017 replication at the NVE eliminates that additional multicast state
1018 in the underlay core, but depending on the multicast traffic
1019 volume, it may cause inefficient use of bandwidth.
1020
1021 <span class="h3"><a class="selflink" id="section-4.2" href="#section-4.2">4.2</a>. Overlay Issues to Consider</span>
1022
1023 <span class="h4"><a class="selflink" id="section-4.2.1" href="#section-4.2.1">4.2.1</a>. Data Plane vs. Control Plane Driven</span>
1024
1025 In the case of an L2 NVE, it is possible to dynamically learn MAC
1026 addresses against VAPs. It is also possible that such addresses be
1027 known and controlled via management or a control protocol for both L2
1028 NVEs and L3 NVEs. Dynamic data-plane learning implies that flooding
1029 of unknown destinations be supported and hence implies that broadcast
1030 and/or multicast be supported or that ingress replication be used as
1031 described in <a href="#section-4.2.3">Section 4.2.3</a>. Multicasting in the underlay network for
1032 dynamic learning may lead to significant scalability limitations.
1033 Specific forwarding rules must be enforced to prevent loops from
1034 happening. This can be achieved using a spanning tree, a shortest
1035 path tree, or a split-horizon mesh.
1036
1037 It should be noted that the amount of state to be distributed is
1038 dependent upon network topology and the number of virtual machines.
1039 Different forms of caching can also be utilized to minimize state
1040 distribution between the various elements. The control plane should
1041 not require an NVE to maintain the locations of all the Tenant
1042 Systems whose VNs are not present on the NVE. The use of a control
1043 plane does not imply that the data plane on NVEs has to maintain all
1044 the forwarding state in the control plane.
1045
1046 <span class="h4"><a class="selflink" id="section-4.2.2" href="#section-4.2.2">4.2.2</a>. Coordination between Data Plane and Control Plane</span>
1047
1048 For an L2 NVE, the NVE needs to be able to determine MAC addresses of
1049 the Tenant Systems connected via a VAP. This can be achieved via
1050 data-plane learning or a control plane. For an L3 NVE, the NVE needs
1051 to be able to determine the IP addresses of the Tenant Systems
1052 connected via a VAP.
1053
1054
1055
1056
1057
1058
1059
1060 <span class="grey">Lasserre, et al. Informational [Page 19]</span></pre>
1061 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-20" ></span>
1062 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
1063
1064
1065 In both cases, coordination with the NVE control protocol is needed
1066 such that when the NVE determines that the set of addresses behind a
1067 VAP has changed, it triggers the NVE control plane to distribute this
1068 information to its peers.
1069
1070 <span class="h4"><a class="selflink" id="section-4.2.3" href="#section-4.2.3">4.2.3</a>. Handling Broadcast, Unknown Unicast, and Multicast (BUM) Traffic</span>
1071
1072 There are several options to support packet replication needed for
1073 broadcast, unknown unicast, and multicast. Typical methods include:
1074
1075 - Ingress replication
1076
1077 - Use of underlay multicast trees
1078
1079 There is a bandwidth vs. state trade-off between the two approaches.
1080 Depending upon the degree of replication required (i.e., the number
1081 of hosts per group) and the amount of multicast state to maintain,
1082 trading bandwidth for state should be considered.
1083
1084 When the number of hosts per group is large, the use of underlay
1085 multicast trees may be more appropriate. When the number of hosts is
1086 small (e.g., 2-3) and/or the amount of multicast traffic is small,
1087 ingress replication may not be an issue.
1088
1089 Depending upon the size of the data center network and hence the
1090 number of (S,G) entries, and also the duration of multicast flows,
1091 the use of underlay multicast trees can be a challenge.
1092
1093 When flows are well known, it is possible to pre-provision such
1094 multicast trees. However, it is often difficult to predict
1095 application flows ahead of time; hence, programming of (S,G) entries
1096 for short-lived flows could be impractical.
1097
1098 A possible trade-off is to use in the underlay shared multicast trees
1099 as opposed to dedicated multicast trees.
1100
1101 <span class="h4"><a class="selflink" id="section-4.2.4" href="#section-4.2.4">4.2.4</a>. Path MTU</span>
1102
1103 When using overlay tunneling, an outer header is added to the
1104 original frame. This can cause the MTU of the path to the egress
1105 tunnel endpoint to be exceeded.
1106
1107 It is usually not desirable to rely on IP fragmentation for
1108 performance reasons. Ideally, the interface MTU as seen by a Tenant
1109 System is adjusted such that no fragmentation is needed.
1110
1111
1112
1113
1114
1115
1116 <span class="grey">Lasserre, et al. Informational [Page 20]</span></pre>
1117 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-21" ></span>
1118 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
1119
1120
1121 It is possible for the MTU to be configured manually or to be
1122 discovered dynamically. Various Path MTU discovery techniques exist
1123 in order to determine the proper MTU size to use:
1124
1125 - Classical ICMP-based Path MTU Discovery [<a href="./rfc1191" title="&quot;Path MTU discovery&quot;">RFC1191</a>] [<a href="./rfc1981" title="&quot;Path MTU Discovery for IP version 6&quot;">RFC1981</a>]
1126
1127 Tenant Systems rely on ICMP messages to discover the MTU of the
1128 end-to-end path to its destination. This method is not always
1129 possible, such as when traversing middleboxes (e.g., firewalls)
1130 that disable ICMP for security reasons.
1131
1132 - Extended Path MTU Discovery techniques such as those defined in
1133 [<a href="./rfc4821" title="&quot;Packetization Layer Path MTU Discovery&quot;">RFC4821</a>]
1134
1135 Tenant Systems send probe packets of different sizes and rely on
1136 confirmation of receipt or lack thereof from receivers to allow a
1137 sender to discover the MTU of the end-to-end paths.
1138
1139 While it could also be possible to rely on the NVE to perform
1140 segmentation and reassembly operations without relying on the Tenant
1141 Systems to know about the end-to-end MTU, this would lead to
1142 undesired performance and congestion issues as well as significantly
1143 increase the complexity of hardware NVEs required for buffering and
1144 reassembly logic.
1145
1146 Preferably, the underlay network should be designed in such a way
1147 that the MTU can accommodate the extra tunneling and possibly
1148 additional NVO3 header encapsulation overhead.
1149
1150 <span class="h4"><a class="selflink" id="section-4.2.5" href="#section-4.2.5">4.2.5</a>. NVE Location Trade-Offs</span>
1151
1152 In the case of DC traffic, traffic originated from a VM is native
1153 Ethernet traffic. This traffic can be switched by a local virtual
1154 switch or ToR switch and then by a DC gateway. The NVE function can
1155 be embedded within any of these elements.
1156
1157 There are several criteria to consider when deciding where the NVE
1158 function should happen:
1159
1160 - Processing and memory requirements
1161
1162 o Datapath (e.g., lookups, filtering, and
1163 encapsulation/decapsulation)
1164
1165 o Control-plane processing (e.g., routing, signaling, and OAM)
1166 and where specific control-plane functions should be enabled
1167
1168 - FIB/RIB size
1169
1170
1171
1172 <span class="grey">Lasserre, et al. Informational [Page 21]</span></pre>
1173 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-22" ></span>
1174 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
1175
1176
1177 - Multicast support
1178
1179 o Routing/signaling protocols
1180
1181 o Packet replication capability
1182
1183 o Multicast FIB
1184
1185 - Fragmentation support
1186
1187 - QoS support (e.g., marking, policing, and queuing)
1188
1189 - Resiliency
1190
1191 <span class="h4"><a class="selflink" id="section-4.2.6" href="#section-4.2.6">4.2.6</a>. Interaction between Network Overlays and Underlays</span>
1192
1193 When multiple overlays co-exist on top of a common underlay network,
1194 resources (e.g., bandwidth) should be provisioned to ensure that
1195 traffic from overlays can be accommodated and QoS objectives can be
1196 met. Overlays can have partially overlapping paths (nodes and
1197 links).
1198
1199 Each overlay is selfish by nature. It sends traffic so as to
1200 optimize its own performance without considering the impact on other
1201 overlays, unless the underlay paths are traffic engineered on a per-
1202 overlay basis to avoid congestion of underlay resources.
1203
1204 Better visibility between overlays and underlays, or general
1205 coordination in placing overlay demands on an underlay network, may
1206 be achieved by providing mechanisms to exchange performance and
1207 liveliness information between the underlay and overlay(s) or by the
1208 use of such information by a coordination system. Such information
1209 may include:
1210
1211 - Performance metrics (throughput, delay, loss, jitter) such as
1212 defined in [<a href="./rfc3148" title="&quot;A Framework for Defining Empirical Bulk Transfer Capacity Metrics&quot;">RFC3148</a>], [<a href="./rfc2679" title="&quot;A One-way Delay Metric for IPPM&quot;">RFC2679</a>], [<a href="./rfc2680" title="&quot;A One-way Packet Loss Metric for IPPM&quot;">RFC2680</a>], and [<a href="./rfc3393" title="&quot;IP Packet Delay Variation Metric for IP Performance Metrics (IPPM)&quot;">RFC3393</a>].
1213
1214 - Cost metrics
1215
1216 <span class="h2"><a class="selflink" id="section-5" href="#section-5">5</a>. Security Considerations</span>
1217
1218 There are three points of view when considering security for NVO3.
1219 First, the service offered by a service provider via NVO3 technology
1220 to a tenant must meet the mutually agreed security requirements.
1221 Second, a network implementing NVO3 must be able to trust the virtual
1222 network identity associated with packets received from a tenant.
1223 Third, an NVO3 network must consider the security associated with
1224 running as an overlay across the underlay network.
1225
1226
1227
1228 <span class="grey">Lasserre, et al. Informational [Page 22]</span></pre>
1229 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-23" ></span>
1230 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
1231
1232
1233 To meet a tenant's security requirements, the NVO3 service must
1234 deliver packets from the tenant to the indicated destination(s) in
1235 the overlay network and external networks. The NVO3 service provides
1236 data confidentiality through data separation. The use of both VNIs
1237 and tunneling of tenant traffic by NVEs ensures that NVO3 data is
1238 kept in a separate context and thus separated from other tenant
1239 traffic. The infrastructure supporting an NVO3 service (e.g.,
1240 management systems, NVEs, NVAs, and intermediate underlay networks)
1241 should be limited to authorized access so that data integrity can be
1242 expected. If a tenant requires that its data be confidential, then
1243 the Tenant System may choose to encrypt its data before transmission
1244 into the NVO3 service.
1245
1246 An NVO3 service must be able to verify the VNI received on a packet
1247 from the tenant. To ensure this, not only tenant data but also NVO3
1248 control data must be secured (e.g., control traffic between NVAs and
1249 NVEs, between NVAs, and between NVEs). Since NVEs and NVAs play a
1250 central role in NVO3, it is critical that secure access to NVEs and
1251 NVAs be ensured such that no unauthorized access is possible. As
1252 discussed in <a href="#section-3.1.5.2">Section 3.1.5.2</a>, identification of Tenant Systems is
1253 based upon state that is often provided by management systems (e.g.,
1254 a VM orchestration system in a virtualized environment). Secure
1255 access to such management systems must also be ensured. When an NVE
1256 receives data from a Tenant System, the tenant identity needs to be
1257 verified in order to guarantee that it is authorized to access the
1258 corresponding VN. This can be achieved by identifying incoming
1259 packets against specific VAPs in some cases. In other circumstances,
1260 authentication may be necessary. Once this verification is done, the
1261 packet is allowed into the NVO3 overlay, and no integrity protection
1262 is provided on the overlay packet encapsulation (e.g., the VNI,
1263 destination NVE, etc.).
1264
1265 Since an NVO3 service can run across diverse underlay networks, when
1266 the underlay network is not trusted to provide at least data
1267 integrity, data encryption is needed to assure correct packet
1268 delivery.
1269
1270 It is also desirable to restrict the types of information (e.g.,
1271 topology information as discussed in <a href="#section-4.2.6">Section 4.2.6</a>) that can be
1272 exchanged between an NVO3 service and underlay networks based upon
1273 their agreed security requirements.
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284 <span class="grey">Lasserre, et al. Informational [Page 23]</span></pre>
1285 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-24" ></span>
1286 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
1287
1288
1289 <span class="h2"><a class="selflink" id="section-6" href="#section-6">6</a>. Informative References</span>
1290
1291 [<a id="ref-EVPN">EVPN</a>] Sajassi, A., Aggarwal, R., Bitar, N., Isaac, A., and J.
1292 Uttaro, "BGP MPLS Based Ethernet VPN", Work in Progress,
1293 <a href="./draft-ietf-l2vpn-evpn-10">draft-ietf-l2vpn-evpn-10</a>, October 2014.
1294
1295 [<a id="ref-OPPSEC">OPPSEC</a>] Dukhovni, V. "Opportunistic Security: Some Protection Most
1296 of the Time", Work in Progress, <a href="./draft-dukhovni-opportunistic-security-04">draft-dukhovni-</a>
1297 <a href="./draft-dukhovni-opportunistic-security-04">opportunistic-security-04</a>, August 2014.
1298
1299 [<a id="ref-RFC1191">RFC1191</a>] Mogul, J. and S. Deering, "Path MTU discovery", <a href="./rfc1191">RFC 1191</a>,
1300 November 1990, &lt;<a href="http://www.rfc-editor.org/info/rfc1191">http://www.rfc-editor.org/info/rfc1191</a>&gt;.
1301
1302 [<a id="ref-RFC1981">RFC1981</a>] McCann, J., Deering, S., and J. Mogul, "Path MTU Discovery
1303 for IP version 6", <a href="./rfc1981">RFC 1981</a>, August 1996,
1304 &lt;<a href="http://www.rfc-editor.org/info/rfc1981">http://www.rfc-editor.org/info/rfc1981</a>&gt;.
1305
1306 [<a id="ref-RFC2679">RFC2679</a>] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way
1307 Delay Metric for IPPM", <a href="./rfc2679">RFC 2679</a>, September 1999,
1308 &lt;<a href="http://www.rfc-editor.org/info/rfc2679">http://www.rfc-editor.org/info/rfc2679</a>&gt;.
1309
1310 [<a id="ref-RFC2680">RFC2680</a>] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way
1311 Packet Loss Metric for IPPM", <a href="./rfc2680">RFC 2680</a>, September 1999,
1312 &lt;<a href="http://www.rfc-editor.org/info/rfc2680">http://www.rfc-editor.org/info/rfc2680</a>&gt;.
1313
1314 [<a id="ref-RFC3148">RFC3148</a>] Mathis, M. and M. Allman, "A Framework for Defining
1315 Empirical Bulk Transfer Capacity Metrics", <a href="./rfc3148">RFC 3148</a>, July
1316 2001, &lt;<a href="http://www.rfc-editor.org/info/rfc3148">http://www.rfc-editor.org/info/rfc3148</a>&gt;.
1317
1318 [<a id="ref-RFC3393">RFC3393</a>] Demichelis, C. and P. Chimento, "IP Packet Delay Variation
1319 Metric for IP Performance Metrics (IPPM)", <a href="./rfc3393">RFC 3393</a>,
1320 November 2002, &lt;<a href="http://www.rfc-editor.org/info/rfc3393">http://www.rfc-editor.org/info/rfc3393</a>&gt;.
1321
1322 [<a id="ref-RFC4364">RFC4364</a>] Rosen, E. and Y. Rekhter, "BGP/MPLS IP Virtual Private
1323 Networks (VPNs)", <a href="./rfc4364">RFC 4364</a>, February 2006,
1324 &lt;<a href="http://www.rfc-editor.org/info/rfc4364">http://www.rfc-editor.org/info/rfc4364</a>&gt;.
1325
1326 [<a id="ref-RFC4761">RFC4761</a>] Kompella, K., Ed., and Y. Rekhter, Ed., "Virtual Private
1327 LAN Service (VPLS) Using BGP for Auto-Discovery and
1328 Signaling", <a href="./rfc4761">RFC 4761</a>, January 2007,
1329 &lt;<a href="http://www.rfc-editor.org/info/rfc4761">http://www.rfc-editor.org/info/rfc4761</a>&gt;.
1330
1331 [<a id="ref-RFC4762">RFC4762</a>] Lasserre, M., Ed., and V. Kompella, Ed., "Virtual Private
1332 LAN Service (VPLS) Using Label Distribution Protocol (LDP)
1333 Signaling", <a href="./rfc4762">RFC 4762</a>, January 2007,
1334 &lt;<a href="http://www.rfc-editor.org/info/rfc4762">http://www.rfc-editor.org/info/rfc4762</a>&gt;.
1335
1336
1337
1338
1339
1340 <span class="grey">Lasserre, et al. Informational [Page 24]</span></pre>
1341 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-25" ></span>
1342 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
1343
1344
1345 [<a id="ref-RFC4821">RFC4821</a>] Mathis, M. and J. Heffner, "Packetization Layer Path MTU
1346 Discovery", <a href="./rfc4821">RFC 4821</a>, March 2007,
1347 &lt;<a href="http://www.rfc-editor.org/info/rfc4821">http://www.rfc-editor.org/info/rfc4821</a>&gt;.
1348
1349 [<a id="ref-RFC6820">RFC6820</a>] Narten, T., Karir, M., and I. Foo, "Address Resolution
1350 Problems in Large Data Center Networks", <a href="./rfc6820">RFC 6820</a>, January
1351 2013, &lt;<a href="http://www.rfc-editor.org/info/rfc6820">http://www.rfc-editor.org/info/rfc6820</a>&gt;.
1352
1353 [<a id="ref-RFC7364">RFC7364</a>] Narten, T., Ed., Gray, E., Ed., Black, D., Fang, L.,
1354 Kreeger, L., and M. Napierala, "Problem Statement:
1355 Overlays for Network Virtualization", <a href="./rfc7364">RFC 7364</a>, October
1356 2014, &lt;<a href="http://www.rfc-editor.org/info/rfc7364">http://www.rfc-editor.org/info/rfc7364</a>&gt;.
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396 <span class="grey">Lasserre, et al. Informational [Page 25]</span></pre>
1397 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-26" ></span>
1398 <span class="grey"><a href="./rfc7365">RFC 7365</a> Framework for DC Network Virtualization October 2014</span>
1399
1400
1401 Acknowledgments
1402
1403 In addition to the authors, the following people contributed to this
1404 document: Dimitrios Stiliadis, Rotem Salomonovitch, Lucy Yong, Thomas
1405 Narten, Larry Kreeger, and David Black.
1406
1407 Authors' Addresses
1408
1409 Marc Lasserre
1410 Alcatel-Lucent
1411 EMail: [email protected]
1412
1413
1414 Florin Balus
1415 Alcatel-Lucent
1416 777 E. Middlefield Road
1417 Mountain View, CA 94043
1418 United States
1419 EMail: [email protected]
1420
1421
1422 Thomas Morin
1423 Orange
1424 EMail: [email protected]
1425
1426
1427 Nabil Bitar
1428 Verizon
1429 50 Sylvan Road
1430 Waltham, MA 02145
1431 United States
1432 EMail: [email protected]
1433
1434
1435 Yakov Rekhter
1436 Juniper
1437 EMail: [email protected]
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452 Lasserre, et al. Informational [Page 26]
1453 </pre>