--- 1/draft-ietf-idr-bgp-analysis-01.txt 2006-02-04 23:29:24.000000000 +0100 +++ 2/draft-ietf-idr-bgp-analysis-02.txt 2006-02-04 23:29:24.000000000 +0100 @@ -1,18 +1,18 @@ INTERNET-DRAFT David Meyer -draft-ietf-idr-bgp-analysis-01.txt Keyur Patel +draft-ietf-idr-bgp-analysis-02.txt Keyur Patel Category Informational Expires: October 2003 April 2003 BGP-4 Protocol Analysis - + Status of this Document This document is an Internet-Draft and is in full conformance with all provisions of Section 10 of RFC2026. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. @@ -58,22 +58,22 @@ 4. BGP Persistent Peer Oscillations . . . . . . . . . . . . . . . 8 5. BGP Performance characteristics and Scalability. . . . . . . . 8 5.1. Link bandwidth and CPU utilization. . . . . . . . . . . . . 8 5.1.1. CPU utilization. . . . . . . . . . . . . . . . . . . . . 9 5.1.2. Memory requirements. . . . . . . . . . . . . . . . . . . 11 6. BGP Policy Expressiveness and its Implications . . . . . . . . 12 6.1. Existence of Unique Stable Routings . . . . . . . . . . . . 13 6.2. Existence of Stable Routings. . . . . . . . . . . . . . . . 14 7. Applicability. . . . . . . . . . . . . . . . . . . . . . . . . 15 8. Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . 16 - 9. References . . . . . . . . . . . . . . . . . . . . . . . . . . 17 - 10. Author's Address. . . . . . . . . . . . . . . . . . . . . . . 18 + 9. Informative References . . . . . . . . . . . . . . . . . . . . 17 + 10. Author's Addresses. . . . . . . . . . . . . . . . . . . . . . 18 11. Full Copyright Statement. . . . . . . . . . . . . . . . . . . 18 1. Introduction BGP-4 is an inter-autonomous system routing protocol designed for TCP/IP internets. Version 1 of the BGP protocol was published in RFC 1105 [RFC1105]. Since then BGP versions 2, 3, and 4 have been developed. Version 2 was documented in RFC 1163 [RFC1163]. Version 3 is documented in RFC 1267 [RFC1267]. Version 4 is documented in the [BGP4] (version 4 of BGP will hereafter be referred to as BGP). The @@ -105,27 +105,27 @@ The key features of the protocol are the notion of path attributes and aggregation of network layer reachability information (NLRI). Path attributes provide BGP with flexibility and extensibility. Path attributes are partitioned into well-known and optional. The provision for optional attributes allows experimentation that may involve a group of BGP routers without affecting the rest of the Internet. New optional attributes can be added to the protocol in much the same way that new options are added to, say, the Telnet protocol [RFC854]. - One of the most important path attributes is the AS_PATH. AS - reachability information traverses the Internet, this information is - augmented by the list of autonomous systems that have been traversed - thus far, forming the AS_PATH. The AS_PATH allows straightforward - suppression of the looping of routing information. In addition, the - AS_PATH serves as a powerful and versatile mechanism for policy-based - routing. + One of the most important path attributes is the Autonomous System + Path, or AS_PATH. AS reachability information traverses the Internet, + this information is augmented by the list of autonomous systems that + have been traversed thus far, forming the AS_PATH. The AS_PATH allows + straightforward suppression of the looping of routing information. In + addition, the AS_PATH serves as a powerful and versatile mechanism + for policy-based routing. BGP enhances the AS_PATH attribute to include sets of autonomous systems as well as lists via the AS_SET attribute. This extended format allows generated aggregate routes to carry path information from the more specific routes used to generate the aggregate. It should be noted however, that as of this writing, AS_SETs are rarely used in the Internet [ROUTEVIEWS]. 2.2. BGP Algorithms @@ -153,21 +153,21 @@ Finally, note that BGP is a self-contained protocol. That is, BGP specifies how routing information is exchanged both between BGP speakers in different autonomous systems, and between BGP speakers within a single autonomous system. 2.3. BGP Finite State Machine (FSM) The BGP FSM is a set of rules that are applied to a BGP speaker's set of configured peers for the BGP operation. A BGP implementation requires that a BGP speaker must connect to and listen on TCP port - 179 for accepting any new BGP connections from it's peers. The BGP + 179 for accepting any new BGP connections from its peers. The BGP Finite State Machine, or FSM, must be initiated and maintained for each new incoming and outgoing peer connections. However, in steady state operation, there will be only one BGP FSM per connection per peer. There may exist a temporary period where in a BGP peer may have separate incoming and outgoing connections resulting into two different BGP FSMs for a peer (instead of one). This can be resolved following BGP connection collision rules defined in the [BGP4]. @@ -185,25 +185,25 @@ OPENSENT: BGP peer is waiting for OPEN message from its peer. OPENCONFIRM: BGP peer is waiting for KEEPALIVE or NOTIFICATION message from its peer. ESTABLISHED: BGP peer connection is established and exchanges UPDATE, NOTIFICATION, and KEEPALIVE messages with its peer. - There are different BGP events that operates on above mentioned states + There are different BGP events that operate on above mentioned states of BGP FSM for its peers. These BGP events are used for initiating and terminating peer connections. They also assist BGP in identifying any - persistent peer connection oscillations and provides mechanism - for controlling it. + persistent peer connection oscillations and provide a mechanism + for controlling them. Following are different BGP events: Manual Start: Manually start the peer connection. Manual Stop: Manually stop the peer connection. Automatic Start: Local system automatically starts the peer connection. @@ -212,70 +212,70 @@ peer connection with peer in passive mode. Automatic start with passive TCP flag: Local system administrator automatically starts the peer connection with peer in passive mode. Automatic start with bgp_stop_flap option set: Local system administrator automatically starts the peer connection with peer oscillation - damping enabled + damping enabled. Automatic start with bgp_stop_flap option set and passive TCP establishment option set: Local system administrator automatically starts the peer connection with peer oscillation damping enabled and with peer in passive mode. Automatic stop: Local system automatically stops the BGP connection. Both, Manual Start and Manual Stop are mandatory BGP events. All other events are optional. 3. BGP Capabilities - The BGP Capability mechanism [RFC2842] provides easy and flexible way - to introduce new features within the protocol. In particular, the BGP - capability mechanism allows peers to negotiate various optional + The BGP Capability mechanism [RFC2842] provides an easy and flexible + way to introduce new features within the protocol. In particular, the + BGP capability mechanism allows peers to negotiate various optional features during startup. This allows the base BGP protocol to contain only essential functionality, while at the same time providing a flexible mechanism for signaling protocol extensions. 4. BGP Persistent Peer Oscillations Ideally, whenever a BGP speaker detects an error in any peer connection, it shuts down the peer and changes its FSM state to IDLE. BGP speaker requires a Start event to re-initiate its idle peer - connection. If error remains persistent and BGP speaker generates + connection. If the error remains persistent and BGP speaker generates Start event automatically then it may result in persistent peer flapping. However, although peer oscillation is found to be wide- spread in BGP implementations, methods for preventing persistent peer oscillations are outside the scope of base BGP protocol specification. 5. BGP Performance characteristics and Scalability In this section, we provide "order of magnitude" answers to the questions of how much link bandwidth, router memory and router CPU cycles the BGP protocol will consume under normal conditions. In particular, we will address the scalability of BGP and its limitations. It is important to note that BGP does not require all the routers within an autonomous system to participate in the BGP protocol. In particular, only the border routers that provide connectivity between the local autonomous system and their adjacent autonomous systems - need participate in BGP. The ability to constraint the set of BGP + need participate in BGP. The ability to constrain the set of BGP speakers is one way to address scaling issues. 5.1. Link bandwidth and CPU utilization Immediately after the initial BGP connection setup, BGP peers exchange complete set of routing information. If we denote the total number of routes in the Internet by N, the mean AS distance of the Internet by M (distance at the level of an autonomous system, expressed in terms of the number of autonomous systems), the total number of autonomous systems in the Internet by A, and assume that @@ -309,21 +309,21 @@ difficult to estimate the actual reduction in bandwidth and processing that BGP has provided over BGP-3. If we simply enumerate all aggregate blocks into their individual class-based networks, we would not take into account "dead" space that has been reserved for future expansion. The best metric for determining the success of BGP's aggregation is to sample the number NLRI entries in the globally connected Internet today and compare it to projected growth rates before BGP was deployed. At the time of this writing, the full set of exterior routes carried - by BGP approximately 120,000 network entries [ROUTEVIEWS]. + by BGP is approximately 120,000 network entries [ROUTEVIEWS]. 5.1.1. CPU utilization An important and fundamental feature of BGP is that BGP's CPU utilization depends only on the stability of the Internet. If the Internet is stable, then the only link bandwidth and router CPU cycles consumed by BGP are due to the exchange of the BGP KEEPALIVE messages. The KEEPALIVE messages are exchanged only between peers. The suggested frequency of the exchange is 30 seconds. The KEEPALIVE messages are quite short (19 octets), and require virtually no @@ -332,50 +332,51 @@ the overhead (in terms of bandwidth and CPU) associated with the KEEPALIVE messages should be viewed as negligible. During periods of Internet instability, changes to the reachability information are passed between routers in UPDATE messages. If we denote the number of routing changes per second by C, then in the worst case the amount of bandwidth consumed by the BGP can be expressed as O(C * M). The greatest overhead per UPDATE message occurs when each UPDATE message contains only a single network. It should be pointed out that in practice routing changes exhibit strong - locality with respect to the AS path. That is routes that change are - likely to have common AS path. In this case multiple networks can be + locality with respect to the AS path. That is, routes that change are + likely to have common AS path. In this case, multiple networks can be grouped into a single UPDATE message, thus significantly reducing the amount of bandwidth required (see also Appendix F.1 of [BGP4]). Since in the steady state the link bandwidth and router CPU cycles consumed by the BGP protocol are dependent only on the stability of the Internet, it follows that BGP should have no scaling problems in the areas of link bandwidth and router CPU utilization. This assumes that as the Internet grows, the overall stability of the inter-AS connectivity of the Internet can be controlled. In particular, while the size of the IPv4 Internet routing table is bounded by O(2^32 * M), (where M is a slow-moving function describing the AS interconnectivity of the network), no such bound can be formulated for the dynamic properties (i.e., stability) of BGP. Finally, since the dynamic properties of the network cannot be quantitatively bounded, stability must be addressed via heuristics such as BGP - Route Flap Dampening [RFC2439]. Due to the nature of BGP, such - dampening should be viewed as a local to an autonomous system matter + Route Flap Damping [RFC2439]. Due to the nature of BGP, such damping + should be viewed as a matter local to an autonomous system matter (see also Appendix F.2 of [BGP4]). It may also be instructive to compare bandwidth and CPU requirements - of BGP with EGP. While with BGP the complete information is exchanged - only at the connection establishment time, with EGP the complete - information is exchanged periodically (usually every 3 minutes). Note - that both for BGP and for EGP the amount of information exchanged is - roughly on the order of the networks reachable via a peer that sends - the information. Therefore, even if one assumes extreme instabilities - of BGP, its worst case behavior will be the same as the steady state - behavior of it's predecessor, EGP. + of BGP with the Exterior Gateway Protocol (EGP). While with BGP the + complete information is exchanged only at the connection + establishment time, with EGP the complete information is exchanged + periodically (usually every 3 minutes). Note that both for BGP and + for EGP the amount of information exchanged is roughly on the order + of the number of networks reachable via a peer that sends the + information. Therefore, even if one assumes extreme instabilities of + BGP, its worst case behavior will be the same as the steady state + behavior of its predecessor, EGP. Operational experience with BGP showed that the incremental update approach employed by BGP provides qualitative improvement in both bandwidth and CPU utilization when compared with complete periodic updates used by EGP (see also presentation by Dennis Ferguson at the Twentieth IETF, March 11-15, 1991, St.Louis). 5.1.2. Memory requirements To quantify the worst case memory requirements for BGP, we denote the @@ -385,33 +386,31 @@ number of autonomous systems in the Internet by A, and the total number of BGP speakers that a system is peering with by K (note that K will usually be dominated by the total number of the BGP speakers within a single autonomous system). Then the worst case memory requirements (MR) can be expressed as MR = O((N + M * A) * K) It is interesting to note that prior to the introduction of BGP in the NSFNET Backbone, memory requirements on the NSFNET Backbone - routers running EGP were on the order of O(N *K). Therefore, the - extra overhead in memory incurred by modern routers running BGP is - less than 7 percent. + routers running EGP were on the order of O(N *K). Since a mean AS distance M is a slow moving function of the interconnectivity ("meshiness") of the Internet, for all practical purposes the worst case router memory requirements are on the order of the total number of networks in the Internet times the number of peers the local system is peering with. We expect that the total number of networks in the Internet will grow much faster than the - average number of peers per router. As a result, BGP's memory - scaling properties are linearly related to the total number of - networks in the Internet. + average number of peers per router. As a result, BGP's memory scaling + properties are linearly related to the total number of networks in + the Internet. The following table illustrates typical memory requirements of a router running BGP. It is assumed that each network is encoded as four bytes, each AS is encoded as two bytes, and each networks is reachable via some fraction of all of the peers (# BGP peers/per net). For purposes of the estimates here, we will calculate MR = ((N*4) + (M*A)*2) * K. # Networks Mean AS Distance # AS's # BGP peers/per net Memory Req (MR) ---------- ---------------- ------ ------------------- -------------- @@ -412,42 +411,43 @@ reachable via some fraction of all of the peers (# BGP peers/per net). For purposes of the estimates here, we will calculate MR = ((N*4) + (M*A)*2) * K. # Networks Mean AS Distance # AS's # BGP peers/per net Memory Req (MR) ---------- ---------------- ------ ------------------- -------------- 100,000 20 3,000 20 1,040,000 100,000 20 15,000 20 1,040,000 120,000 10 15,000 100 75,000,000 140,000 15 20,000 100 116,000,000 + In analyzing BGP's memory requirements, we focus on the size of the forwarding table (and ignoring implementation details). In particular, we derive upper bounds for the size of the forwarding table. For example, at the time of this writing, the forwarding - tables of a typical backbone router carries on the order of 120,000 + tables of a typical backbone router carry on the order of 120,000 entries. Given this number, one might ask whether it would be possible to have a functional router with a table that will have 1,000,000 entries. Clearly the answer to this question is independent of BGP. Interestingly, in his review of the BGP protocol for the BGP review committee in March of 1990, Paul Tsuchiya noted that "BGP does not scale well. This is not really the fault of BGP. It is the fault of the flat IP address space. Given the flat IP address space, any routing protocol must carry network numbers in its updates." The - introduction of the provider based aggregation schemes (e.g., CIDR - [RFC1519]) have sought to address this issue, to the extent possible, - within the context of current addressing architectures. + introduction of the provider based aggregation schemes (e.g., RFC + 1519 [RFC1519]) have sought to address this issue, to the extent + possible, within the context of current addressing architectures. 6. BGP Policy Expressiveness and its Implications BGP is unique among deployed IP routing protocols in that routing is - determined using semantically rich routing policies. Although - routing policies are usually the first thing that comes to a network + determined using semantically rich routing policies. Although routing + policies are usually the first thing that comes to a network operator's mind concerning BGP, it is important to note that the languages and techniques for specifying BGP routing policies are not actually a part of the protocol specification (see RFC 2622 [RFC2622] for an example of such a policy language). In addition, the BGP specification contains few restrictions, either explicitly or implicitly, on routing policy languages. These languages have typically been developed by vendors and have evolved through interactions with network engineers in an environment lacking vendor- independent standards. @@ -494,22 +494,22 @@ AS1 ------> AS2 /|\ | | | | | | \|/ AS3 ------- AS4 That is, the AS3-AS4 link is intended to be used only when the AS2-AS4 link is down (for outbound traffic, AS4 simply gives routes from AS2 a higher local preference). This is a common scenario in - today's Internet. But note that this configuration has another - stable solution: + today's Internet. But note that this configuration has another stable + solution: AS1 ------- AS2 | | | | | | \|/ \|/ AS3 ------> AS4 In this case, AS3 does not translate the "depref my route" community received from AS4 into a "depref my route" community for AS1, and so @@ -533,26 +533,27 @@ possible with RFC 1997 communities. At the same time, applications of communities by network operators are evolving to address complex issues of inter-domain traffic engineering. 6.2. Existence of Stable Routings One can also construct a set of policies for which BGP can not guarantee that a stable routing exists (or worse, that a stable routing will ever be found). For example, RFC 3345 [RFC3345] documents several scenarios that lead to route oscillations - associated with the use of MEDs. Route oscillation will happen in BGP - when a set of policies has no solution. That is, when there is no - stable routing that satisfies the constraints imposed by policy, then - BGP has no choice by to keep trying. In addition, BGP configurations - can have a stable routing, yet the protocol may not be able to find - it; BGP can "get trapped" down a blind alley that has no solution. + associated with the use of the Multi-Exit Discriminator or MED, + attribute. Route oscillation will happen in BGP when a set of + policies has no solution. That is, when there is no stable routing + that satisfies the constraints imposed by policy, then BGP has no + choice by to keep trying. In addition, BGP configurations can have a + stable routing, yet the protocol may not be able to find it; BGP can + "get trapped" down a blind alley that has no solution. Protocol divergence is not, however, a problem associated solely with use of the MED attribute. This potential exists in BGP even without the use of the MED attribute. Hence, like the unintended nondeterminism described in the previous section, this type of protocol divergence is an unintended consequence of the unconstrained nature of BGP policy languages. 7. Applicability @@ -587,37 +588,37 @@ mechanisms. To summarize, BGP is well suitable as an inter-autonomous system routing protocol for the IPv4 Internet that is based on IP [RFC791] as the Internet Protocol and "hop-by-hop" routing paradigm. Finally, BGP is equally applicable to IPv6 [RFC2460] internets. 8. Acknowledgments We would like to thank Paul Traina for authoring previous versions of - this document. Tim Griffin also provided many insightful comments on - earlier versions of this document. + this document. Tim Griffin and Randy Presuhn also provided many + insightful comments on earlier versions of this document. -9. References +9. Informative References - [BGP4] Rekhter, Y., T. Li., and Hares. S, Editors, "A + [BGP4] Rekhter, Y., T. Li., and S. Hares, Editors, "A Border Gateway Protocol 4 (BGP-4)", draft-ietf-idr-bgp4-19.txt. Work in progress. [RFC791] "INTERNET PROTOCOL", DARPA INTERNET PROGRAM PROTOCOL SPECIFICATION, RFC 791, September, 1981. - [RFC854] Postel, J. and Reynolds, J., "TELNET PROTOCOL + [RFC854] Postel, J. and J. Reynolds, "TELNET PROTOCOL SPECIFICATION", RFC 854, May, 1983. - [RFC1105] Lougheed, K., and Rekhter, Y, "Border Gateway + [RFC1105] Lougheed, K., and Y. Rekhter, "Border Gateway Protocol BGP", RFC 1105, June 1989. [RFC1163] Lougheed, K., and Rekhter, Y, "Border Gateway Protocol BGP", RFC 1105, June 1990. [RFC1264] Hinden, R., "Internet Routing Protocol Standardization Criteria", RFC 1264, October 1991. [RFC1267] Lougheed, K., and Rekhter, Y, "Border Gateway Protocol 3 (BGP-3)", RFC 1105, October 1991. @@ -630,44 +631,44 @@ [RFC1771] Rekhter, Y., and T. Li, "A Border Gateway Protocol 4 (BGP-4)", RFC 1771, March 1995. [RFC1772] Rekhter, Y., and P. Gross, Editors, "Application of the Border Gateway Protocol in the Internet", RFC 1772, March 1995. [RFC1997] Chandra. R, and T. Li, "BGP Communities Attribute", RFC 1997, August, 1996. - [RFC2439] Villamizar, C., Chandra, R., and Govindan, R., + [RFC2439] Villamizar, C., Chandra, R., and R. Govindan, "BGP Route Flap Damping", RFC 2439, November 1998. [RFC2460] Deering, S, and R. Hinden, "Internet Protocol, Version 6 (IPv6) Specification", RFC 2460, December, 1998. [RFC2622] C. Alaettinoglu, et al., "Routing Policy Specification Language (RPSL)" RFC 2622, May, 1998. [RFC2842] Chandra, R. and J. Scudder, "Capabilities Advertisement with BGP-4", RFC 2842, May 2000. [RFC3345] McPherson, D., Gill, V., Walton, D., and - Retana, A, "Border Gateway Protocol (BGP) Persistent + A. Retana, "Border Gateway Protocol (BGP) Persistent Route Oscillation Condition", RFC 3345, August, 2002. - [ROUTEVIEWS] Meyer, David, "The Route Views Project", + [ROUTEVIEWS] Meyer, D., "The Route Views Project", http://www.routeviews.org -10. Author's Address +10. Author's Addresses David Meyer Email: dmm@maoz.com Keyur Patel Cisco Systems Email: keyupate@cisco.com 11. Full Copyright Statement