| 1 |
<pre>Network Working Group T. Bates |
| 2 |
Request for Comments: 2260 Cisco Systems |
| 3 |
Category: Informational Y. Rekhter |
| 4 |
Cisco Systems |
| 5 |
January 1998 |
| 6 |
|
| 7 |
|
| 8 |
<span class="h1">Scalable Support for Multi-homed Multi-provider Connectivity</span> |
| 9 |
|
| 10 |
Status of this Memo |
| 11 |
|
| 12 |
This memo provides information for the Internet community. It does |
| 13 |
not specify an Internet standard of any kind. Distribution of this |
| 14 |
memo is unlimited. |
| 15 |
|
| 16 |
Copyright Notice |
| 17 |
|
| 18 |
Copyright (C) The Internet Society (1998). All Rights Reserved. |
| 19 |
|
| 20 |
<span class="h2"><a class="selflink" id="section-2" href="#section-2">2</a>. Abstract</span> |
| 21 |
|
| 22 |
This document describes addressing and routing strategies for multi- |
| 23 |
homed enterprises attached to multiple Internet Service Providers |
| 24 |
(ISPs) that are intended to reduce the routing overhead due to these |
| 25 |
enterprises in the global Internet routing system. |
| 26 |
|
| 27 |
<span class="h2"><a class="selflink" id="section-3" href="#section-3">3</a>. Motivations</span> |
| 28 |
|
| 29 |
An enterprise may acquire its Internet connectivity from more than |
| 30 |
one Internet Service Provider (ISP) for some of the following |
| 31 |
reasons. Maintaining connectivity via more than one ISP could be |
| 32 |
viewed as a way to make connectivity to the Internet more reliable. |
| 33 |
This way when connectivity through one of the ISPs fails, |
| 34 |
connectivity via the other ISP(s) would enable the enterprise to |
| 35 |
preserve its connectivity to the Internet. In addition to providing |
| 36 |
more reliable connectivity, maintaining connectivity via more than |
| 37 |
one ISP could also allow the enterprise to distribute load among |
| 38 |
multiple connections. For enterprises that span wide geographical |
| 39 |
area this could also enable better (more optimal) routing. |
| 40 |
|
| 41 |
The above considerations, combined with the decreasing prices for the |
| 42 |
Internet connectivity, motivate more and more enterprises to become |
| 43 |
multi-homed to multiple ISPs. At the same time, the routing overhead |
| 44 |
that such enterprises impose on the Internet routing system becomes |
| 45 |
more and more significant. Scaling the Internet, and being able to |
| 46 |
support a growing number of such enterprises demands mechanism(s) to |
| 47 |
contain this overhead. This document assumes that an approach where |
| 48 |
routers in the "default-free" zone of the Internet would be required |
| 49 |
|
| 50 |
|
| 51 |
|
| 52 |
<span class="grey">Bates & Rekhter Informational [Page 1]</span></pre> |
| 53 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span> |
| 54 |
<span class="grey"><a href="./rfc2260">RFC 2260</a> Multihoming January 1998</span> |
| 55 |
|
| 56 |
|
| 57 |
to maintain a route for every multi-homed enterprise that is |
| 58 |
connected to multiple ISPs does not provide an adequate scaling. |
| 59 |
Moreover, given the nature of the Internet, this document assumes |
| 60 |
that any approach to handle routing for such enterprises should |
| 61 |
minimize the amount of coordination among ISPs, and especially the |
| 62 |
ISPs that are not directly connected to these enterprises. |
| 63 |
|
| 64 |
There is a difference of opinions on whether the driving factors |
| 65 |
behind multi-homing to multiple ISPs could be adequately addressed by |
| 66 |
multi-homing just to a single ISP, which would in turn eliminate the |
| 67 |
negative impact of multi-homing on the Internet routing system. |
| 68 |
Discussion of this topic is beyond the scope of this document. |
| 69 |
|
| 70 |
The focus of this document is on the routing and addressing |
| 71 |
strategies that could reduce the routing overhead due to multi-homed |
| 72 |
enterprises connected to multiple ISPs in the Internet routing |
| 73 |
system. |
| 74 |
|
| 75 |
The strategies described in this document are equally applicable to |
| 76 |
both IPv4 and IPv6. |
| 77 |
|
| 78 |
<span class="h2"><a class="selflink" id="section-4" href="#section-4">4</a>. Address allocation and assignment</span> |
| 79 |
|
| 80 |
A multi-homed enterprise connected to a set of ISPs would be |
| 81 |
allocated a block of addresses (address prefix) by each of these ISPs |
| 82 |
(an enterprise connected to N ISPs would get N different blocks). |
| 83 |
The address allocation from the ISPs to the enterprise would be based |
| 84 |
on the "address-lending" policy [<a href="./rfc2008">RFC2008</a>]. The allocated addresses |
| 85 |
then would be used for address assignment within the enterprise. |
| 86 |
|
| 87 |
One possible address assignment plan that the enterprise could employ |
| 88 |
is to use the topological proximity of a node (host) to a particular |
| 89 |
ISP (to the interconnect between the enterprise and the ISP) as a |
| 90 |
criteria for selecting which of the address prefixes to use for |
| 91 |
address assignment to the node. A particular node (host) may be |
| 92 |
assigned address(es) out of a single prefix, or may have addresses |
| 93 |
from different prefixes. |
| 94 |
|
| 95 |
<span class="h2"><a class="selflink" id="section-5" href="#section-5">5</a>. Routing information exchange</span> |
| 96 |
|
| 97 |
The issue of routing information exchange between an enterprise and |
| 98 |
its ISPs is decomposed into the following components: |
| 99 |
|
| 100 |
a) reachability information that an enterprise border router |
| 101 |
advertises to a border router within an ISP |
| 102 |
|
| 103 |
b) reachability information that a border router within an ISP |
| 104 |
advertises to an enterprise border router |
| 105 |
|
| 106 |
|
| 107 |
|
| 108 |
<span class="grey">Bates & Rekhter Informational [Page 2]</span></pre> |
| 109 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span> |
| 110 |
<span class="grey"><a href="./rfc2260">RFC 2260</a> Multihoming January 1998</span> |
| 111 |
|
| 112 |
|
| 113 |
The primary focus of this document is on (a); (b) is covered only as |
| 114 |
needed by this document. |
| 115 |
|
| 116 |
<span class="h3"><a class="selflink" id="section-5.1" href="#section-5.1">5.1</a>. Advertising reachability information by enterprise border routers</span> |
| 117 |
|
| 118 |
When an enterprise border router connected to a particular ISP |
| 119 |
determines that the connectivity between the enterprise and the |
| 120 |
Internet is up through all of its ISPs, the router advertises (to the |
| 121 |
border router of that ISP) reachability to only the address prefix |
| 122 |
that the ISP allocated to the enterprise. This way in a steady state |
| 123 |
routes injected by the enterprise into its ISPs are aggregated by |
| 124 |
these ISPs, and are not propagated into the "default-free" zone of |
| 125 |
the Internet. |
| 126 |
|
| 127 |
When an enterprise border router connected to a particular ISP |
| 128 |
detemrines that the connectivity between the enterprise and the |
| 129 |
Internet through one or more of its other ISPs is down, the router |
| 130 |
starts advertising reachability to the address prefixes that was |
| 131 |
allocated by these ISPs to the enterprise. This would result in |
| 132 |
injecting additional routing information into the "default-free" zone |
| 133 |
of the Internet. However, one could observe that the probability of |
| 134 |
all multi-homed enterprises in the Internet concurrently losing |
| 135 |
connectivity to the Internet through one or more of their ISPs is |
| 136 |
fairly small. Thus on average the number of additional routes in the |
| 137 |
"default-free" zone of the Internet due to multi-homed enterprises is |
| 138 |
expected to be a small fraction of the total number of such |
| 139 |
enterprises. |
| 140 |
|
| 141 |
The approach described above is predicated on the assumption that an |
| 142 |
enterprise border router has a mechanism(s) by which it could |
| 143 |
determine (a) whether the connectivity to the Internet through some |
| 144 |
other border router of that enterprise is up or down, and (b) the |
| 145 |
address prefix that was allocated to the enterprise by the ISP |
| 146 |
connected to the other border router. One such possible mechanism |
| 147 |
could be provided by BGP [<a href="./rfc1771">RFC1771</a>]. In this case border routers |
| 148 |
within the enterprise would have an IBGP peering with each other. |
| 149 |
Whenever one border router determines that the intersection between |
| 150 |
the set of reachable destinations it receives via its EBGP (from its |
| 151 |
directly connected ISP) peerings and the set of reachable |
| 152 |
destinations it receives from another border router (in the same |
| 153 |
enterprise) via IBGP is empty, the border router would start |
| 154 |
advertising to its external peer reachability to the address prefix |
| 155 |
that was allocated to the enterprise by the ISP connected to the |
| 156 |
other border router. The other border router would advertise (via |
| 157 |
IBGP) the address prefix that was allocated to the enterprise by the |
| 158 |
ISP connected to that router. This approach is known as "auto route |
| 159 |
injection". |
| 160 |
|
| 161 |
|
| 162 |
|
| 163 |
|
| 164 |
<span class="grey">Bates & Rekhter Informational [Page 3]</span></pre> |
| 165 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span> |
| 166 |
<span class="grey"><a href="./rfc2260">RFC 2260</a> Multihoming January 1998</span> |
| 167 |
|
| 168 |
|
| 169 |
As an illustration consider an enterprise connected to two ISPs, |
| 170 |
ISP-A and ISP-B. Denote the enterprise border router that connects |
| 171 |
the enterprise to ISP-A as BR-A; denote the enterprise border router |
| 172 |
that connects the enterprise to ISP-B as BR-B. Denote the address |
| 173 |
prefix that ISP-A allocated to the enterprise as Pref-A; denote the |
| 174 |
address prefix that ISP-B allocated to the enterprise as Pref-B. |
| 175 |
When the set of routes BR-A receives from ISP-A (via EBGP) has a |
| 176 |
non-empty intersection with the set of routes BR-A receives from BR-B |
| 177 |
(via IBGP), BR-A advertises to ISP-A only the reachability to Pref-A. |
| 178 |
When the intersection becomes empty, BR-A would advertise to ISP-A |
| 179 |
reachability to both Pref-A and Pref-B. This would continue for as |
| 180 |
long as the intersection remains empty. Once the intersection becomes |
| 181 |
non-empty, BR-A would stop advertising reachability to Pref-B to |
| 182 |
ISP-A (but would still continue to advertise reachability to Pref-A |
| 183 |
to ISP-A). Figure 1 below describes this method graphically. |
| 184 |
|
| 185 |
+-------+ +-------+ +-------+ +-------+ |
| 186 |
( ) ( ) ( ) ( ) |
| 187 |
( ISP-A ) ( ISP-B ) ( ISP-A ) ( ISP-B ) |
| 188 |
( ) ( ) ( ) ( ) |
| 189 |
+-------+ +-------+ +-------+ +-------+ |
| 190 |
| /\ | /\ | /\ | |
| 191 |
| || | || | Pref-A (connection |
| 192 |
| Pref-A | Pref-B | Pref-B broken) |
| 193 |
| || | || | || | |
| 194 |
+-----+ +-----+ +-----+ +-----+ |
| 195 |
| BR-A|------|BR-B | | BR-A|------|BR-B | |
| 196 |
+-----+ IBGP +-----+ +-----+ IBGP +-----+ |
| 197 |
|
| 198 |
non-empty intersection empty intersection |
| 199 |
|
| 200 |
|
| 201 |
Figure 1: Reachability information advertised |
| 202 |
|
| 203 |
Although strictly an implementation detail, calculating the |
| 204 |
intersection could potentially be a costly operation for a large set |
| 205 |
of routes. An alternate solution to this is to make use of a selected |
| 206 |
single (or more) address prefix received from an ISP (the ISP's |
| 207 |
backbone route for example) and configure the enterprise border |
| 208 |
router to perform auto route injection if the selected prefix is not |
| 209 |
present via IBGP. Let's suppose ISP-B has a well known address |
| 210 |
prefix, ISP-Pref-B for its backbone. ISP-B advertises this to BR-B |
| 211 |
and BR-B in turn advertises this via IBGP to BR-A. If BR-A sees a |
| 212 |
withdraw for ISP-Pref-B it advertises Pref-B to ISP-A. |
| 213 |
|
| 214 |
|
| 215 |
|
| 216 |
|
| 217 |
|
| 218 |
|
| 219 |
|
| 220 |
<span class="grey">Bates & Rekhter Informational [Page 4]</span></pre> |
| 221 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span> |
| 222 |
<span class="grey"><a href="./rfc2260">RFC 2260</a> Multihoming January 1998</span> |
| 223 |
|
| 224 |
|
| 225 |
The approach described in this section may produce less than the full |
| 226 |
Internet-wide connectivity in the presence of ISPs that filter out |
| 227 |
routes based on the length of their address prefixes. One could |
| 228 |
observe however, that this would be a problem regardless of how the |
| 229 |
enterprise would set up its routing and addressing. |
| 230 |
|
| 231 |
<span class="h3"><a class="selflink" id="section-5.2" href="#section-5.2">5.2</a>. Further improvements</span> |
| 232 |
|
| 233 |
The approach described in the previous section allows to |
| 234 |
significantly reduce the routing overhead in the "default-free" zone |
| 235 |
of the Internet due to multi-homed enterprises. The approach |
| 236 |
described in this section allows to completely eliminate this |
| 237 |
overhead. |
| 238 |
|
| 239 |
An enterprise border router would maintain EBGP peering not just with |
| 240 |
the directly connected border router of an ISP, but with the border |
| 241 |
router(s) in one or more ISPs that have their border routers directly |
| 242 |
connected to the other border routers within the enterprise. We |
| 243 |
refer to such peering as "non-direct" EBGP. |
| 244 |
|
| 245 |
An ISP that maintains both direct and non-direct EBGP peering with a |
| 246 |
particular enterprise would advertise the same set of routes over |
| 247 |
both of these peerings. An enterprise border router that maintains |
| 248 |
either direct or non-direct peering with an ISP advertises to that |
| 249 |
ISP reachability to the address prefix that was allocated by that ISP |
| 250 |
to the enterprise. Within the ISP routes received over direct |
| 251 |
peering should be preferred over routes received over non-direct |
| 252 |
peering. Likewise, within the enterprise routes received over direct |
| 253 |
peering should be preferred over routes received over non-direct |
| 254 |
peering. |
| 255 |
|
| 256 |
Forwarding along a route received over non-direct peering should be |
| 257 |
accomplished via encapsulation [<a href="./rfc1773">RFC1773</a>]. |
| 258 |
|
| 259 |
As an illustration consider an enterprise connected to two ISPs, |
| 260 |
ISP-A and ISP-B. Denote the enterprise border router that connects |
| 261 |
the enterprise to ISP-A as E-BR-A, and the ISP-A border router that |
| 262 |
is connected to E-BR-A as ISP-BR-A; denote the enterprise border |
| 263 |
router that connects the enterprise to ISP-B as E-BR-B, and the ISP-B |
| 264 |
border router that is connected to E-BR-B as ISP-BR-B. Denote the |
| 265 |
address prefix that ISP-A allocated to the enterprise as Pref-A; |
| 266 |
denote the address prefix that ISP-B allocated to the enterprise as |
| 267 |
Pref-B. E-BR-A maintains direct EBGP peering with ISP-BR-A and |
| 268 |
advertises reachability to Pref-A over that peering. E-BR-A also |
| 269 |
maintain a non-direct EBGP peering with ISP-BR-B and advertises |
| 270 |
reachability to Pref-B over that peering. E-BR-B maintains direct |
| 271 |
EBGP peering with ISP-BR-B, and advertises reachability to Pref-B |
| 272 |
over that peering. E-BR-B also maintains a non-direct EBGP peering |
| 273 |
|
| 274 |
|
| 275 |
|
| 276 |
<span class="grey">Bates & Rekhter Informational [Page 5]</span></pre> |
| 277 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-6" ></span> |
| 278 |
<span class="grey"><a href="./rfc2260">RFC 2260</a> Multihoming January 1998</span> |
| 279 |
|
| 280 |
|
| 281 |
with ISP-BR-A, and advertises reachability to Pref-A over that |
| 282 |
peering. |
| 283 |
|
| 284 |
When connectivity between the enterprise and both of its ISPs (ISP-A |
| 285 |
and ISP-B is up, traffic destined to hosts whose addresses were |
| 286 |
assigned out of Pref-A would flow through ISP-A to ISP-BR-A to E-BR- |
| 287 |
A, and then into the enterprise. Likewise, traffic destined to hosts |
| 288 |
whose addresses were assigned out of Pref-B would flow through ISP-B |
| 289 |
to ISP-BR-B to E-BR-B, and then into the enterprise. Now consider |
| 290 |
what would happen when connectivity between ISP-BR-B and E-BR-B goes |
| 291 |
down. In this case traffic to hosts whose addresses were assigned |
| 292 |
out of Pref-A would be handled as before. But traffic to hosts whose |
| 293 |
addresses were assigned out of Pref-B would flow through ISP-B to |
| 294 |
ISP-BR-B, ISP-BR-B would encapsulate this traffic and send it to E- |
| 295 |
BR-A, where the traffic will get decapsulated and then be sent into |
| 296 |
the enterprise. Figure 2 below describes this approach graphically. |
| 297 |
|
| 298 |
+---------+ +---------+ |
| 299 |
( ) ( ) |
| 300 |
( ISP-A ) ( ISP-B ) |
| 301 |
( ) ( ) |
| 302 |
+---------+ +---------+ |
| 303 |
| | |
| 304 |
+--------+ +--------+ |
| 305 |
|ISP-BR-A| |ISP-BR-B| |
| 306 |
+--------+ +--------+ |
| 307 |
| /+/ | |
| 308 |
/\ | Pref-B /+/ | |
| 309 |
|| | /+/ \./ |
| 310 |
Pref-A| /+/ non- /.\ |
| 311 |
|| | /+/ direct | |
| 312 |
| /+/ EBGP | |
| 313 |
+------+ +-------+ |
| 314 |
|E-BR-A|-----------|E-BR-B | |
| 315 |
+------+ IBGP +-------+ |
| 316 |
|
| 317 |
|
| 318 |
Figure 2: Reachability information advertised via non-direct EBGP |
| 319 |
|
| 320 |
Observe that with this scheme there is no additional routing |
| 321 |
information due to multi-homed enterprises that has to be carried in |
| 322 |
the "default-free" zone of the Internet. In addition this scheme |
| 323 |
doesn't degrade in the presence of ISPs that filter out routes based |
| 324 |
on the length of their address prefixes. |
| 325 |
|
| 326 |
Note that the set of routers within an ISP that maintain non-direct |
| 327 |
peering with the border routers within an enterprise doesn't have to |
| 328 |
be restricted to the ISP's border routers that have direct peering |
| 329 |
|
| 330 |
|
| 331 |
|
| 332 |
<span class="grey">Bates & Rekhter Informational [Page 6]</span></pre> |
| 333 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-7" ></span> |
| 334 |
<span class="grey"><a href="./rfc2260">RFC 2260</a> Multihoming January 1998</span> |
| 335 |
|
| 336 |
|
| 337 |
with the enterprise's border routers. The non-direct peering could be |
| 338 |
maintained with any router within the ISP. Doing this could improve |
| 339 |
the overall robustness in the presence of failures within the ISP. |
| 340 |
|
| 341 |
<span class="h3"><a class="selflink" id="section-5.3" href="#section-5.3">5.3</a>. Combining the two</span> |
| 342 |
|
| 343 |
One could observe that while the approach described in <a href="#section-5.2">Section 5.2</a> |
| 344 |
allows to completely eliminate the routing overhead due to multi- |
| 345 |
homed enterprises in the "default-free" zone of the Internet, it may |
| 346 |
result in a suboptimal routing in the presence of link failures. The |
| 347 |
sub-optimality could be reduced by combining the approach described |
| 348 |
in <a href="#section-5.2">Section 5.2</a> with a slightly modified version of the approach |
| 349 |
described in <a href="#section-5.1">Section 5.1</a>. The modification consists of constraining |
| 350 |
the scope of propagation of additional routes that are advertised by |
| 351 |
an enterprise border router when the router detects problems with the |
| 352 |
Internet connectivity through its other border routers. A way to |
| 353 |
constrain the scope is by using the BGP Community attribute |
| 354 |
[<a href="./rfc1997">RFC1997</a>]. |
| 355 |
|
| 356 |
<span class="h3"><a class="selflink" id="section-5.4" href="#section-5.4">5.4</a>. Better (more optimal) routing in steady state</span> |
| 357 |
|
| 358 |
The approach described in this document assumes that in a steady |
| 359 |
state an enterprise border router would advertise to a directly |
| 360 |
connected ISP border router only the reachability to the address |
| 361 |
prefix that this ISP allocated to the enterprise. As a result, |
| 362 |
traffic originated by other enterprises connected to that ISP and |
| 363 |
destined to the parts of the enterprise numbered out of other address |
| 364 |
prefixes would not enter the enterprise at this border router, |
| 365 |
resulting in potentially suboptimal paths. To improve the situation |
| 366 |
the border router may (in steady state) advertise reachability not |
| 367 |
only to the address prefix that was allocated by the ISP that the |
| 368 |
router is directly connected to, but to the address prefixes |
| 369 |
allocated by some other ISPs (directly connected to some other border |
| 370 |
routers within the enterprise). Distribution of such advertisements |
| 371 |
should be carefully constrained, or otherwise this may result in |
| 372 |
significant additional routing information that would need to be |
| 373 |
maintained in the "default-free" part of the Internet. A way to |
| 374 |
constrain the distribution of such advertisements is by using the BGP |
| 375 |
Community attribute [<a href="./rfc1997">RFC1997</a>]. |
| 376 |
|
| 377 |
<span class="h2"><a class="selflink" id="section-6" href="#section-6">6</a>. Comparison with other approaches</span> |
| 378 |
|
| 379 |
CIDR [<a href="./rfc1518">RFC1518</a>] proposes several possible address allocation |
| 380 |
strategies for multi-homed enterprises that are connected to multiple |
| 381 |
ISPs. The following briefly reviews the alternatives being used |
| 382 |
today, and compares them with the approaches described above. |
| 383 |
|
| 384 |
|
| 385 |
|
| 386 |
|
| 387 |
|
| 388 |
<span class="grey">Bates & Rekhter Informational [Page 7]</span></pre> |
| 389 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-8" ></span> |
| 390 |
<span class="grey"><a href="./rfc2260">RFC 2260</a> Multihoming January 1998</span> |
| 391 |
|
| 392 |
|
| 393 |
<span class="h3"><a class="selflink" id="section-6.1" href="#section-6.1">6.1</a>. Solution 1</span> |
| 394 |
|
| 395 |
One possible solution suggested in [<a href="./rfc1518">RFC1518</a>] is for each multi-homed |
| 396 |
enterprise to obtain its IP address space independently from the ISPs |
| 397 |
to which it is attached. This allows each multi-homed enterprise to |
| 398 |
base its IP assignments on a single prefix, and to thereby summarize |
| 399 |
the set of all IP addresses reachable within that enterprise via a |
| 400 |
single prefix. The disadvantage of this approach is that since the |
| 401 |
IP address for that enterprise has no relationship to the addresses |
| 402 |
of any particular ISPs, the reachability information advertised by |
| 403 |
the enterprise is not aggregatable with any, but default route. |
| 404 |
results in the routing overhead in the "default-free" zone of the |
| 405 |
Internet of O(N), where N is the total number of multi-homed |
| 406 |
enterprises across the whole Internet that are connected to multiple |
| 407 |
ISPs. |
| 408 |
|
| 409 |
As a result, this approach can't be viewed as a viable alternative |
| 410 |
for all, but the enterprises that provide high enough degree of |
| 411 |
addressing information aggregation. Since by definition the number of |
| 412 |
such enterprises is likely to be fairly small, this approach isn't |
| 413 |
viable for most of the multi-homed enterprises connected to multiple |
| 414 |
ISPs. |
| 415 |
|
| 416 |
<span class="h3"><a class="selflink" id="section-6.2" href="#section-6.2">6.2</a>. Solution 2</span> |
| 417 |
|
| 418 |
Another possible solution suggested in [<a href="./rfc1518">RFC1518</a>] is to assign each |
| 419 |
multi-homed enterprise a single address prefix, based on one of its |
| 420 |
connections to one of its ISPs. Other ISPs to which the multi-homed |
| 421 |
enterprise is attached maintain a routing table entry for the |
| 422 |
organization, but are extremely selective in terms of which other |
| 423 |
ISPs are told of this route and would need to perform "proxy" |
| 424 |
aggregation. Most of the complexity associated with this approach is |
| 425 |
due to the need to perform "proxy" aggregation, which in turn |
| 426 |
requires t addiional inter-ISP coordination and more complex router |
| 427 |
configuration. |
| 428 |
|
| 429 |
<span class="h2"><a class="selflink" id="section-7" href="#section-7">7</a>. Discussion</span> |
| 430 |
|
| 431 |
The approach described in this document assumes that addresses that |
| 432 |
an enterprise would use are allocated based on the "address lending" |
| 433 |
policy. Consequently, whenever an enterprise changes its ISP, the |
| 434 |
enterprise would need to renumber part of its network that was |
| 435 |
numbered out of the address block that the ISP allocated to the |
| 436 |
enterprise. However, these issues are not specific to multihoming |
| 437 |
and should be considered accepted practice in todays internet. The |
| 438 |
approach described in this document effectively eliminates any |
| 439 |
distinction between single-home and multi-homed enterprise with |
| 440 |
respect to the impact of changing ISPs on renumbering. |
| 441 |
|
| 442 |
|
| 443 |
|
| 444 |
<span class="grey">Bates & Rekhter Informational [Page 8]</span></pre> |
| 445 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-9" ></span> |
| 446 |
<span class="grey"><a href="./rfc2260">RFC 2260</a> Multihoming January 1998</span> |
| 447 |
|
| 448 |
|
| 449 |
The approach described in this document also requires careful address |
| 450 |
assignment within an enterprise, as address assignment impacts |
| 451 |
traffic distribution among multiple connections between an enterprise |
| 452 |
and its ISPs. |
| 453 |
|
| 454 |
Both the issue of address assignment and renumbering could be |
| 455 |
addressed by the appropriate use of network address translation |
| 456 |
(NAT). The use of NAT for multi-homed enterprises is the beyond the |
| 457 |
scope of this document. |
| 458 |
|
| 459 |
Use of auto route injection (as described in <a href="#section-5.1">Section 5.1</a>) increases |
| 460 |
the number of routers in the default-free zone of the Internet that |
| 461 |
could be affected by changes in the connectivity of multi-homed |
| 462 |
enterprises, as compared to the use of provider-independed addresses |
| 463 |
(as described in <a href="#section-6.1">Section 6.1</a>). Specifically, with auto route |
| 464 |
injection when a multi-homed enterprise loses its connectivity |
| 465 |
through one of its ISPs, the auto injected route has to be propagated |
| 466 |
to all the routers in the default-free zone of the Internet. In |
| 467 |
contrast, when an enterprise uses provider-independent addresses, |
| 468 |
only some (but not all) of the routers in the default-free zone would |
| 469 |
see changes in routing when the enterprise loses its connectivity |
| 470 |
through one of its ISPs. |
| 471 |
|
| 472 |
To supress excessive routing load due to link flapping the auto |
| 473 |
injected route has to be advertised until the connectivity via the |
| 474 |
other connection (that was previously down and that triggered auto |
| 475 |
route injection) becomes stable. |
| 476 |
|
| 477 |
Use of the non-direct EBGP approach (as described in <a href="#section-5.2">Section 5.2</a>) |
| 478 |
allows to eliminate route flapping due to multi-homed enterprises in |
| 479 |
the default-free zone of the Internet. That is the non-direct EBGP |
| 480 |
approach has better properties with respect to routing stability than |
| 481 |
the use of provider-independent addresses (as described in <a href="#section-6.1">Section</a> |
| 482 |
<a href="#section-6.1">6.1</a>). |
| 483 |
|
| 484 |
<span class="h2"><a class="selflink" id="section-8" href="#section-8">8</a>. Applications to multi-homed ISPs</span> |
| 485 |
|
| 486 |
The approach described in this document could be applicable to a |
| 487 |
small to medium size ISP that is connected to several upstream ISPs. |
| 488 |
The ISP would acquire blocks of addresses (address prefixes) from its |
| 489 |
upstream ISPs, and would use these addresses for allocations to its |
| 490 |
customers. Either auto route injection, or the non-direct EBGP |
| 491 |
approach, or a combination of both could be used by the ISP when |
| 492 |
peering with its upstream ISPs. Doing this would provide routability |
| 493 |
for the customers of such ISP, without advertsely affecting the |
| 494 |
overall scalability of the Internet routing system. |
| 495 |
|
| 496 |
|
| 497 |
|
| 498 |
|
| 499 |
|
| 500 |
<span class="grey">Bates & Rekhter Informational [Page 9]</span></pre> |
| 501 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-10" ></span> |
| 502 |
<span class="grey"><a href="./rfc2260">RFC 2260</a> Multihoming January 1998</span> |
| 503 |
|
| 504 |
|
| 505 |
<span class="h2"><a class="selflink" id="section-9" href="#section-9">9</a>. Security Considerations</span> |
| 506 |
|
| 507 |
Since the non-direct EBGP approach (as described in <a href="#section-5.2">Section 5.2</a>) |
| 508 |
requires EBGP sessions between routers that are more than one IP hop |
| 509 |
from each other, routers that maintain these sessions should use an |
| 510 |
appropriate authentication mechanism(s) for BGP peer authentication. |
| 511 |
|
| 512 |
Security issues related to the IBGP peering, as well as the EBGP |
| 513 |
peering between routers that are one IP hop from each other are |
| 514 |
outside the scope of this document. |
| 515 |
|
| 516 |
<span class="h2"><a class="selflink" id="section-10" href="#section-10">10</a>. Acknowledgments</span> |
| 517 |
|
| 518 |
The authors of this document do not make any claims on the |
| 519 |
originality of the ideas described in this document. Anyone who |
| 520 |
thought about these ideas before should be given all due credit. |
| 521 |
|
| 522 |
<span class="h2"><a class="selflink" id="section-11" href="#section-11">11</a>. References</span> |
| 523 |
|
| 524 |
[<a id="ref-RFC1518">RFC1518</a>] |
| 525 |
Rekhter, Y., and T. Li, "An Architecture for IP Address |
| 526 |
Allocation with CIDR", <a href="./rfc1518">RFC 1518</a>, September 1993. |
| 527 |
|
| 528 |
[<a id="ref-RFC1771">RFC1771</a>] |
| 529 |
Rekhter, Y., and T. Li, "A Border Gateway Protocol 4 (BGP-4)", |
| 530 |
<a href="./rfc1771">RFC 1771</a>, March 1995. |
| 531 |
|
| 532 |
[<a id="ref-RFC1773">RFC1773</a>] |
| 533 |
Hanks, S., Li, T., Farinacci, T., and P. Traina, "Generic |
| 534 |
Routing Encapsulation over IPv4 networks", <a href="./rfc1773">RFC 1773</a>, October |
| 535 |
1994. |
| 536 |
|
| 537 |
[<a id="ref-RFC1918">RFC1918</a>] |
| 538 |
Rekhter, Y., Moskowitz, B., Karrenberg, D., de Groot G.J., and |
| 539 |
E. Lear, "Address Allocation for Private Internets", <a href="./rfc1918">RFC 1918</a>, |
| 540 |
February 1996. |
| 541 |
|
| 542 |
[<a id="ref-RFC1997">RFC1997</a>] |
| 543 |
Chandra, R., Traina, P., and T. Li, "BGP Communities Attribute", |
| 544 |
<a href="./rfc1997">RFC 1997</a>, August 1996. |
| 545 |
|
| 546 |
[<a id="ref-RFC2008">RFC2008</a>] |
| 547 |
Rekhter, Y., and T. Li, "Implications of Various Address |
| 548 |
Allocation Policies for Internet Routing", <a href="https://www.rfc-editor.org/bcp/bcp7">BCP 7</a>, <a href="./rfc2008">RFC 2008</a>, |
| 549 |
October 1996. |
| 550 |
|
| 551 |
|
| 552 |
|
| 553 |
|
| 554 |
|
| 555 |
|
| 556 |
<span class="grey">Bates & Rekhter Informational [Page 10]</span></pre> |
| 557 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-11" ></span> |
| 558 |
<span class="grey"><a href="./rfc2260">RFC 2260</a> Multihoming January 1998</span> |
| 559 |
|
| 560 |
|
| 561 |
<span class="h2"><a class="selflink" id="section-12" href="#section-12">12</a>. Authors' Addresses</span> |
| 562 |
|
| 563 |
Tony Bates |
| 564 |
Cisco Systems, Inc. |
| 565 |
170 West Tasman Drive |
| 566 |
San Jose, CA 95134 |
| 567 |
|
| 568 |
EMail: [email protected] |
| 569 |
|
| 570 |
|
| 571 |
Yakov Rekhter |
| 572 |
Cisco Systems, Inc. |
| 573 |
170 West Tasman Drive |
| 574 |
San Jose, CA 95134 |
| 575 |
EMail: [email protected] |
| 576 |
|
| 577 |
|
| 578 |
|
| 579 |
|
| 580 |
|
| 581 |
|
| 582 |
|
| 583 |
|
| 584 |
|
| 585 |
|
| 586 |
|
| 587 |
|
| 588 |
|
| 589 |
|
| 590 |
|
| 591 |
|
| 592 |
|
| 593 |
|
| 594 |
|
| 595 |
|
| 596 |
|
| 597 |
|
| 598 |
|
| 599 |
|
| 600 |
|
| 601 |
|
| 602 |
|
| 603 |
|
| 604 |
|
| 605 |
|
| 606 |
|
| 607 |
|
| 608 |
|
| 609 |
|
| 610 |
|
| 611 |
|
| 612 |
<span class="grey">Bates & Rekhter Informational [Page 11]</span></pre> |
| 613 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-12" ></span> |
| 614 |
<span class="grey"><a href="./rfc2260">RFC 2260</a> Multihoming January 1998</span> |
| 615 |
|
| 616 |
|
| 617 |
<span class="h2"><a class="selflink" id="section-13" href="#section-13">13</a>. Full Copyright Statement</span> |
| 618 |
|
| 619 |
Copyright (C) The Internet Society (1998). All Rights Reserved. |
| 620 |
|
| 621 |
This document and translations of it may be copied and furnished to |
| 622 |
others, and derivative works that comment on or otherwise explain it |
| 623 |
or assist in its implementation may be prepared, copied, published |
| 624 |
and distributed, in whole or in part, without restriction of any |
| 625 |
kind, provided that the above copyright notice and this paragraph are |
| 626 |
included on all such copies and derivative works. However, this |
| 627 |
document itself may not be modified in any way, such as by removing |
| 628 |
the copyright notice or references to the Internet Society or other |
| 629 |
Internet organizations, except as needed for the purpose of |
| 630 |
developing Internet standards in which case the procedures for |
| 631 |
copyrights defined in the Internet Standards process must be |
| 632 |
followed, or as required to translate it into languages other than |
| 633 |
English. |
| 634 |
|
| 635 |
The limited permissions granted above are perpetual and will not be |
| 636 |
revoked by the Internet Society or its successors or assigns. |
| 637 |
|
| 638 |
This document and the information contained herein is provided on an |
| 639 |
"AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING |
| 640 |
TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING |
| 641 |
BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION |
| 642 |
HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF |
| 643 |
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. |
| 644 |
|
| 645 |
|
| 646 |
|
| 647 |
|
| 648 |
|
| 649 |
|
| 650 |
|
| 651 |
|
| 652 |
|
| 653 |
|
| 654 |
|
| 655 |
|
| 656 |
|
| 657 |
|
| 658 |
|
| 659 |
|
| 660 |
|
| 661 |
|
| 662 |
|
| 663 |
|
| 664 |
|
| 665 |
|
| 666 |
|
| 667 |
|
| 668 |
Bates & Rekhter Informational [Page 12] |
| 669 |
</pre> |