Inter-Domain Routing T. Li Internet-Draft Cisco Systems, Inc. Intended status: Standards Track G. Huston Expires: December 15, 2007 APNIC June 13, 2007 BGP Stability Improvements draft-li-bgp-stability-01 Status of this Memo By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79. 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. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt. The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. This Internet-Draft will expire on December 15, 2007. Copyright Notice Copyright (C) The IETF Trust (2007). Abstract BGP is the routing protocol used to tie the Autonomous Systems (ASes) of the Internet together. The ongoing stability of BGP in the face of arbitrary inputs, both malicious and unintentional, is of primary importance to the overall stability of the Internet. The overall issue is not a new one. Previously, one aspect of stability, known as route flap damping was originally discussed in RFC 2439. In the intervening years, a great deal of experience with flap damping and Li & Huston Expires December 15, 2007 [Page 1] Internet-Draft BGP Stability Improvements June 2007 other stability concerns has been accumulated. Most recently, the issue of BGP stability has been highlighted in RAWS. This document describes the experience that has been gained concerning stability in the intervening years, hypotheses about remaining problems, suggestions for experiments to be performed, and proposals for possible alternatives. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.1. Requirements Language . . . . . . . . . . . . . . . . . . 3 1.2. History . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3. Observations . . . . . . . . . . . . . . . . . . . . . . . 4 1.3.1. Path hunting . . . . . . . . . . . . . . . . . . . . . 4 1.4. The wave model . . . . . . . . . . . . . . . . . . . . . . 5 1.4.1. Refraction and diffraction . . . . . . . . . . . . . . 5 2. Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.1. Flap damping . . . . . . . . . . . . . . . . . . . . . . . 6 2.2. Rapid convergence . . . . . . . . . . . . . . . . . . . . 6 2.3. Reduced overhead . . . . . . . . . . . . . . . . . . . . . 7 3. Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . 7 3.1. Turn it off . . . . . . . . . . . . . . . . . . . . . . . 7 3.2. Alternate parameters . . . . . . . . . . . . . . . . . . . 7 3.3. Band-stop filtering . . . . . . . . . . . . . . . . . . . 8 3.4. Path length damping . . . . . . . . . . . . . . . . . . . 8 3.5. Optimal path hysteresis . . . . . . . . . . . . . . . . . 10 3.5.1. Optimal path length hysteresis . . . . . . . . . . . . 11 3.6. Delayed best path selection . . . . . . . . . . . . . . . 11 3.7. Deprecate MRAI . . . . . . . . . . . . . . . . . . . . . . 12 3.8. Hybrid approaches . . . . . . . . . . . . . . . . . . . . 13 3.9. Other work . . . . . . . . . . . . . . . . . . . . . . . . 13 4. Next steps . . . . . . . . . . . . . . . . . . . . . . . . . . 13 4.1. Call for collaboration . . . . . . . . . . . . . . . . . . 13 4.2. Literature search . . . . . . . . . . . . . . . . . . . . 13 4.3. Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 14 4.3.1. A Case Study: Path length damping . . . . . . . . . . 14 4.4. Prototyping, Testing and Pilot Deployment . . . . . . . . 19 5. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 19 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 19 7. Security Considerations . . . . . . . . . . . . . . . . . . . 19 8. References . . . . . . . . . . . . . . . . . . . . . . . . . . 20 8.1. Normative References . . . . . . . . . . . . . . . . . . . 20 8.2. Informative References . . . . . . . . . . . . . . . . . . 20 8.3. Potential References . . . . . . . . . . . . . . . . . . . 20 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 21 Intellectual Property and Copyright Statements . . . . . . . . . . 23 Li & Huston Expires December 15, 2007 [Page 2] Internet-Draft BGP Stability Improvements June 2007 1. Introduction BGP [RFC4271] is the routing protocol used to tie the Autonomous Systems (ASes) of the Internet together. The ongoing stability of BGP in the face of arbitrary inputs, both malicious and unintentional, is of primary importance to the overall stability of the Internet. The overall issue is not a new one. Previously, one aspect of stability, known as route flap damping was originally discussed in RFC 2439 [RFC2439]. In the intervening years, a great deal of experience with flap damping and other stability concerns has been accumulated. Most recently, the issue of BGP stability has been highlighted in RAWS [I-D.iab-raws-report]. This document describes the experience that has been gained concerning stability in the intervening years, hypotheses about remaining problems, suggestions for experiments to be performed, and proposals for possible alternatives. Please note that this document is very much a work-in-progress. 1.1. Requirements Language The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 [RFC2119]. 1.2. History The circuits used in computer networks have the unfortunate property that they can intermittently fail and then recover. This was an especially common failure mode for copper-based circuits. Under such circumstances, when there was a BGP speaker on both ends of the circuit, any prefixes advertised across the link would tend to oscillate at the frequency induced by the intermittent link. The oscillating prefixes would then propagate across the full Internet, causing the entire routing subsystem to churn at the rate of the prefix. Individually, a single such prefix is not a significant issue. However, as the Internet continued to scale upwards, it became obvious that the CPU requirements to deal with the ever-increasing number of oscillating prefixes would quickly become onerous. This was aggravated by the fact that the party responsible for the flapping circuit was frequently unaware of the problem, or, worse yet, unwilling to address the issue. Route flap can also be caused by policy oscillations [CN2000] or MED churn oscillations [RFC3345]. Li & Huston Expires December 15, 2007 [Page 3] Internet-Draft BGP Stability Improvements June 2007 Thus, the original goal of route flap damping was to protect the control plane from oscillations. This was done by determining the number of flaps and the time elapsed since the last transition. This is fed into an exponential decay function, and, if the prefix is found to be flapping based on this data, the actual propagation of the route is suppressed. Since the frequency information must be stored even if the prefix is not currently active, there is state overhead associated with flap damping for each prefix that has been oscillating. 1.3. Observations Unfortunately, flap damping isn't truly discerning about the nature of routing changes. Any routing change can easily be misinterpreted by flap damping as instability, resulting in premature damping of prefixes [Harmful]. 1.3.1. Path hunting One source of path changes is BGP's normal mechanism for _path exploration_ or _path hunting_. These situations occur because BGP is a path-vector protocol, where each BGP speaker advertises the path that it is using to its neighbors, complete with the full AS path to the destination. Since the number of possible paths through even a simple topology is large, there can be many different path transitions that can possibly be advertised. Path hunting can occur both when a prefix is first advertised or when a prefix is withdrawn. At advertisement time, the prefix may propagate through the topology at different rates, sometimes resulting in it first appearing at an AS with a suboptimal path. Over time, optimal paths will appear where suboptimal paths were before, resulting in a path change that is subsequently propagated. Similarly, when a prefix is withdrawn from the network, as the local BGP speaker receives the prefix withdrawal from a BGP peer it may substitute a previously received announcement for this prefix from a different peer in its place and announce this replacement route to its peers in response to the received withdrawal. If this replacement path is subsequently withdrawn, the local BGP speaker will again select another previously received announcement from a different peer, if one exists. This process may continue until the original prefix withdrawal has propagated across the entire routing space. Interestingly, the amount of path hunting can increase dramatically as the meshiness of the topology increases. It's easy to observe this if you first consider an acyclic topology (i.e., a tree). In Li & Huston Expires December 15, 2007 [Page 4] Internet-Draft BGP Stability Improvements June 2007 such a topology, there is only one possible path, so there is no hunting. If a single link is added to this topology, then there is one cycle in the graph and at most two possible paths for BGP to explore. Subsequent links can add many more alternate paths, depending on their placement. In the worst possible case path hunting in BGP can explore every possible path of each path length. More commonly it has been observed that path hunting in today's Internet can add an additional 2 or 3 BGP updates to a prefix withdrawal. 1.4. The wave model An intuitive means of understanding the observed behavior is by analogy to a wave. Any change in the network triggers the dissemination of information (either updates or withdrawals) through the topology from the point of occurrence. The information travels outwards along all of the paths supported by BGP, in much the same way that a wave would propagate from a pebble dropped in a lake. The wave expands at each BGP speaker, where the information is propagated to all other BGP peers, including ones that already have the information. If the newly arrived information is inferior to the existing path information, then the wave dies out at that BGP speaker. If the newly arrived path is the best path, then the wave continues to propagate. 1.4.1. Refraction and diffraction Information does not traverse the BGP mesh at constant rates. Differences in implementations, processing loads, propagation delay, damping parameters, and policy can all contribute to delaying updates. As with a physical wave, we know that the speed of propagation varies with the material. This results in a bending of the wave, known as _refraction_. Similarly, since the BGP mesh is not a continuous medium we get _diffraction_, where the wave passes through an aperture and fans out from there, creating a new wavefront. Each additional wavefront represents additional processing burden on the routing subsystem. It is interesting to note that flap damping itself may be a contributor to the creation of additional wavefronts. Since a route that is being damped will be delayed for a long time, damping is effectively delaying a wave of information, possibly creating more refractive and diffractive effects. Another major contributor to the generation of additional Li & Huston Expires December 15, 2007 [Page 5] Internet-Draft BGP Stability Improvements June 2007 intermediate wavefronts of information is the disparate use of the Minimum Route Advertisement Interval (MRAI) Timer, where various implementations impose differing delays on update propagation. It should be noted that the wave analogy does break down when it comes to interference. Unlike a physical wave, two waves of BGP information do not interfere additively or destructively. Instead, as noted above, at most one wave will propagate from any point, and which wave will propagate may vary from point to point. 2. Goals 2.1. Flap damping As we reconsider the mechanisms that constitute flap damping, we need to keep in mind that the original goals of detecting and protecting the routing subsystem from noisy inputs is still a requirement. While copper circuits are now less common in the core, the overall network has expanded dramatically and there is a wide variance in the skills and experience in operational roles. As a result, it is still possible for an errant AS to inject flapping information into the BGP mesh, either as the result of policy misconfiguration, implementation error, an intermittent circuit, or even as an intentional destructive act. Thus, it is important that there still be mechanisms that intervene and ameliorate these effects, protecting the routing subsystem. 2.2. Rapid convergence While protecting the routing system is of paramount importance, it is vital that the routing subsystem also continue to perform its primary task: providing routing. Any flap damping and path hunting suppression mechanism must continue to provide rapid convergence to some workable path so that connectivity is restored. However, this goal should not be construed to require rapid optimality. While a best path should eventually be selected and propagated, it is far more important that some connectivity be restored immediately. Most applications can survive with a sub-optimal path, while no applications can succeed if no path is selected. This distinction seems vital for understanding the precise behavior of any mechanism, so for the sake of this discussion, we explicitly define two milestones for convergence: Li & Huston Expires December 15, 2007 [Page 6] Internet-Draft BGP Stability Improvements June 2007 reachability: The source has a valid path to the destination. The path may be suboptimal and may not respect the ultimate traffic engineering preferences of the ASes along the path. As a result, the path may exhibit congestion, unwelcome QoS handling, or any other of a number of sub-optimalities. Time-sensitive applications such as video or voice may require the optimal path for proper operation. optimality: The source has the long-term best path to the destination. This milestone has been reached when there is a stable transitive closure of the best path selection process of all ASes along the path from source to destination. 2.3. Reduced overhead Any form of damping of BGP updates should strive to remove the unnecessary exchange of information. As described above, both path hunting and refractive effects cause unnecessary churn in BGP. The flap damping mechanism should be generalized to help suppress as much of this unnecessary information as possible. 3. Hypotheses In this section, we put forth a number of hypotheses about possible mechanisms to achieve the goals above. As of this writing, more investigation is needed on each of these theories, and where possible we've included some discussion of the experiments that we feel would be worthwhile. Our goal here is to examine a number of different mechanisms, understand their relative benefits, and select a small subset to become the core set of replacement mechanisms. 3.1. Turn it off Given the concern about the negative effects of path damping, [RIPE-378] recommends that path damping be disabled. While this is not unreasonable given the lack of beneficial alternatives, we feel that some of the possibilities presented here will eventually prevail and that this sentiment can be changed over time. 3.2. Alternate parameters It has been suggested in [Harmful] that the default flap damping parameters in existing implementations are simply too aggressive and quickly convert normal path hunting into a damping event that precludes connectivity. Significantly increasing the parameters could permit significantly more churn to be passed by the routing subsystem while still filtering out truly periodic sources of flap. Li & Huston Expires December 15, 2007 [Page 7] Internet-Draft BGP Stability Improvements June 2007 It would be useful to test this by simply configuring numerous differing parameters and observing if there is any beneficial effect. At this time, we have no recommendations for possible alternative parameter settings. 3.3. Band-stop filtering Another view is that classical flap damping isn't working as well as we might like because of a frequency mismatch. The current mechanism looks for a number of changes in a given period of time. If the route exceeds this threshold frequency, then it is damped. The assumed information model behind the current BGP Flap Damping parameters was that of a relatively low frequency oscillation occurring over an extended period of time. Measurements of BGP activity indicate that one of the predominant signal components is a high frequency oscillation occurring over a short time interval. This acts as a false positive for damping that we would like to avoid. One alternative approach is to shift from looking for flapping above a given threshold frequency and simply accept that when there is a real topological change, there will be extensive high frequency path changes. These should be dealt with by separate path hunting suppression techniques, as described below. Some time after the topological change, the high-frequency path changes should stop and the route should then resume its stability. Subsequent path changes would then be indicative of real oscillation and would be subject to damping. The implementation of this would be relatively straightforward. When a change is seen on a stable route, it opens an oscillation window of a fixed duration (e.g., 60s). Any changes within that window are not considered as contributing to flap damping. After the window is closed, any subsequent changes would count as significant events towards damping. Effectively, this technique creates a filter that passes very, very low frequencies (e.g., configuration changes) and defers high frequency changes to path hunting mechanisms, but will detect and deter ongoing route flap within a certain frequency band. This is normally known as a band-stop filter. 3.4. Path length damping The increased meshiness of the core of the Internet has significantly changed the nature of path changes that are visible in BGP. As the meshiness of the network increases, the number of parallel links between any given pair of ASes tends to increase. This helps protect against single link failures between ASes. This also reduces the frequency of AS path changes on transit prefixes because most of the Li & Huston Expires December 15, 2007 [Page 8] Internet-Draft BGP Stability Improvements June 2007 link failures in the densely meshed part of the network will not result in an AS path change. As a result, when a BGP speaker does see a change in the AS path, and in particular, when the AS path length increases, this would seem to be a good heuristic indication that there is some significant failure. The ultimate trigger for the advertisement of an update with a longer AS path is the removal of a previously used shorter path. This is either due to withdrawal at the origin, or removal of an interior connection. As a result, it seems likely that such failures are less likely to have alternative working paths and that the increase in path length is a harbinger of path hunting that is likely to be unsuccessful. We therefore suggest that this event could be used to trigger a suppression period, which would allow the prefix to oscillate arbitrarily without propagation to the remainder of the network. The obvious risk is that this would be a false negative, unnecessarily disrupting connectivity. Observations of the time-coupling of updates in BGP that refer to the same prefix show that only 28% of all BGP updates occur as an isolated event, 26% of all updates occur as a part of a pair of updates occurring within 65 seconds of each other and the remaining 46% of all BGP Updates occur within a coupled sequence of 3 or more updates where each update occurs within 65 seconds of the previous update in the sequence. Again, the implementation of this would be relatively straightforward. When a BGP speaker found that it needed to change its best path for a prefix and that the new best path was longer than the previous best path, then it would suppress the advertisement of the longer AS path to its neighbor and start a timer. Subsequent changes to the prefix that continue to lengthen the AS path would restart the timer. If the BGP speaker installed a shorter best path, or undertook a local withdrawal it would remove the suppressed update and cancel the timer. Otherwise, when the timer expired, the BGP speaker would advertise the result, if any. Some analysis suggests that this technique will be effective at suppressing about 20% of the overall churn rate. [Potaroo0607] The benefits of this approach are that it would drastically reduce the amount of path hunting that happened when a stub site had a failure of its tail circuit. This approach has a number of drawbacks: 1. Failures that result in alternate best paths with path lengths that are equal to the previous best path are not damped, even in Li & Huston Expires December 15, 2007 [Page 9] Internet-Draft BGP Stability Improvements June 2007 the case of stub tail-circuit failures. 2. Convergence is harmed if the alternate AS paths that are damped out are optimal: A. If the failure that triggered the path change is not due to a tail-circuit failure, then useful alternate AS paths will be ignored. B. While this approach is beneficial for stub sites, such sites are not particularly good candidates for being explicitly routed. Such sites should only ever need prefixes that are aggregateable. Such prefixes may be explicitly distributed by BGP for the sake of traffic engineering, so understanding the scope of affected prefixes is left for future study. For non-stub sites, this approach damages convergence. Unfortunately, it seems difficult to distinguish between stub prefixes and non-stub prefixes. To do so would seem to require require explicit annotation that could only be reasonably generated by manual configuration and would likely be incorrect. Transporting the annotation itself would require further protocol extension. 3.5. Optimal path hysteresis It has been observed that the overall topology of the Internet at the AS level changes at a fairly low rate. Thus, the optimal AS path to a given prefix, ignoring transient issues, changes at a very low rate. This suggests that caching the optimal AS path and waiting for it to reappear would be another alternative heuristic to help select only the long-term optimal path. An implementation of this technique might retain a copy of the AS path on per-prefix basis, even if it had no active path to the prefix. Because most implementations maintain a cache of AS paths, this is not necessarily prohibitively expensive. When a new AS path is received for a prefix, the new path is compared to the cached optimal path. If it matches, or it is preferable to the stored optimal path, then the new path is immediately accepted, advertised, and the cache can be updated appropriately. However, if the new AS path is inferior to the cached path, then the implementation can infer that there is some path hunting in progress and can choose to either not perform best path selection, not select the new path, or not advertise the new path. Again, after a suitable period has elapsed, the implementation may decide that the optimal path is unlikely to appear and may process the inferior path normally. The benefits of this approach are that when a site has been Li & Huston Expires December 15, 2007 [Page 10] Internet-Draft BGP Stability Improvements June 2007 temporarily unreachable for a short time, then when it returns to connectivity, only the optimal path will propagate through the network. There are three drawbacks to this approach. First, this approach will delay convergence in the case where the cached optimal path is not going to be restored. Second, this approach will delay reachability. Third, the maintenance of the cache could be a non- trivial amount of overhead. Many implementations maintain a cache of AS paths, where the actual path is stored a single time and then various prefixes point to a cache entry. In such circumstances, if the AS path is maintained for some other active prefix, then the additional cost of caching the path is the additional entry for the unreachable prefix and the pointer to the cache entry. However, if there are no other references for the AS path, then the storage of the AS path itself would be part of the overhead. The impact of this overhead is tempered by the fact that a prefix that is only disconnected from the network for a short time would likely reuse the same memory shortly thereafter in any case. An implementation could also alleviate the cost of this overhead by limiting the amount of memory spent on caching optimal paths for inactive prefixes, or temporally limiting the lifetime of the cached information. 3.5.1. Optimal path length hysteresis A variant of this approach would be to only cache the path length of the optimal path. This would allow certain suboptimal paths matching the cached length to pass rapidly through the system, and in exchange, it would decrease the amount of overhead necessary to maintain the cache. 3.6. Delayed best path selection Another observation based on the discussion in Section 1.4.1 is that the amount of flap is exacerbated by each AS selecting the best possible path each time a new path is presented. This is not strictly required by BGP, so ignoring some of the incoming paths would be perfectly acceptable. Further, an implementation could reasonably delay performing any best path analysis for an arbitrarily long time, as long as it continued to advertise the path it actually used. Thus, one possible policy would be to only perform best path selection when absolutely required. When the first path for a prefix arrives, the implementation would immediately select that path, thereby restoring connectivity. Subsequent paths from other neighbors for the same prefix would not trigger a new best path computation. Rather, they would simply start a timer that would only expire when the paths had stabilized. Li & Huston Expires December 15, 2007 [Page 11] Internet-Draft BGP Stability Improvements June 2007 The benefit of this approach is that it would provide rapid reachability and a major reduction in churn. By propagating one path and suppressing most of the intermediate paths, only a few paths are likely to be propagated. The drawback to this approach is that it will still necessary propagate some suboptimal paths. If the initial path was the long-term optimal path, then churn would not be an issue. Thus, under this approach, propagating the first path helps to optimize reachability, but makes it very likely that the optimal path will subsequently follow. Further, if the neighbor that relayed the initial path decides to change its path selection for any reason, then the local BGP speaker will have no alternative except to execute best path selection and propagate a new best path. This would likely be another intermediate path, not necessarily the long-term optimal path. Some basic analysis shows that this technique, when combined with optimal path hysteresis is capable of reducing the overall BGP message load by 35% for prefixes that oscillate frequently. In addition, when these techniques are combined with path length damping, there is negligible further improvement. Examination of the result shows that optimal path hysteresis also effectively damps out much of the path hunting messaging that occurs when a prefix is being withdrawn, in much the same way that path length damping would do. 3.7. Deprecate MRAI BGP specifies that a prefix should not be advertised multiple times within a fixed period of time. This is called the Min Route Advertisement Interval (MRAI). Implementations of this are sometimes simplistic and can result in information being delayed for the length of this interval even when such delay causes an increase in the number of updates due to diffractive replications. It also leads to unnecessary delays in convergence. The original intent of MRAI was to rate-limit BGP updates to prevent thrashing. Unfortunately, this unsophisticated control has some side effects that are deleterious to the overall effort. If other can provide appropriate guarantees that the update rate will remain appropriately constrained, then the spirit of the MRAI requirement would be satisfied and the actual mechanism itself would not be necessary. For example, path length damping could be used to ensure that the rate of increases in the AS path length would be kept to a controllable level. Refinements in flap damping might be used to deal with the case where the AS path length is constant. No mechanism would be necessary to deal with a decreasing AS path length, since that is necessarily bounded by the AS path length Li & Huston Expires December 15, 2007 [Page 12] Internet-Draft BGP Stability Improvements June 2007 itself. In such circumstances, it should be possible to remove the MRAI mechanism entirely, improve convergence, decrease diffraction, and continue to ensure overall mesh stability. 3.8. Hybrid approaches The approaches listed here could, in principle, be used in conjunction with one another. This would result in a hybrid behavior that had the benefits and drawbacks of the combined mechanisms. For example, it should be possible to combine optimal path caching and delayed best path selection. This approach would then propagate the first path seen by a BGP speaker, but would then delay subsequent path selection opportunities until the optimal path is seen. 3.9. Other work Other work is already in progress to help reduce the amount of BGP messages that are generated when a large number of routes are withdrawn. [AggrWithdrawl] 4. Next steps 4.1. Call for collaboration As can be seen from the above, there is a great deal of work yet to be done on this subject. Collaborators are most welcome in any aspect of the work. 4.2. Literature search There are a number of technical articles listed below that have been published on BGP flap damping and stability that need to be reviewed and included if they prove substantive. A few known ones are listed here. There are very likely a number of other articles in the literature that are relevant. [TON-1998] [Infocom-1999] [FTCS-1999] [Sigcomm-2000] [Infocom-2001] Li & Huston Expires December 15, 2007 [Page 13] Internet-Draft BGP Stability Improvements June 2007 [Sigcomm-2002] [PCC-2004] [Infocom-2005] [R-BGP] 4.3. Analysis A number of projects have collected traces of BGP update messages that demonstrate both flap and path hunting. It would be of great interest to examine the effects of some of the proposal in Section 3 in detail on these traces. 4.3.1. A Case Study: Path length damping This case study examines the BGP updates over the month of April 2007 based on an observation point adjacent to AS 4637. During that month a single eBGP peering session received 1,341,520 BGP update messages, reflecting 3,523,906 individual prefix updates and 627,538 individual prefix withdrawals. Considering that there were an average of some 215,000 individual prefixes in the BGP routing table across the month, that's an average of around 19 updates for every prefix in the month, assuming a uniform distribution of updates across the entire routing domain. Of course, the distribution of updates is not uniform, and most of the network is highly stable. Half of these 210,000 prefixes had less than 10 routing updates across the month, and only 20,000 prefixes had more than 40 updates for the month. In other words, this is a very skewed distribution, with 10% of announced prefixes being responsible for 53% of all routing updates, and the busiest 1% of prefixes responsible for 24% of the routing updates for the month. The first step is to look at what kinds of updates one can expect from a single peer in BGP. The following table classifies the types of BGP updates of interest here. Li & Huston Expires December 15, 2007 [Page 14] Internet-Draft BGP Stability Improvements June 2007 +------+------------------------------------------------------------+ | Code | Description | +------+------------------------------------------------------------+ | AA+ | Announcement of an already announced prefix with a longer | | | AS path (update to longer path) | | AA- | Announcement of an announced prefix with a shorter AS path | | | (update to shorter path) | | AA0 | Announcement of an announced prefix with a different path | | | of the same length (update to a different AS path of same | | | length) | | AA* | Announcement of an announced prefix with the same path but | | | different attributes (update of attributes) | | AA | Announcement of an announced prefix with no change in path | | | or attributes (possible BGP error or data collection | | | error) | | WA+ | Announcement of a withdrawn prefix, with longer AS path | | WA- | Announcement of a withdrawn prefix, with shorter AS path | | WA0 | Announcement of a withdrawn prefix, with different AS path | | | of the same length | | WA* | Announcement of a withdrawn prefix with the same AS path, | | | but different attributes | | WA | Announcement of a withdrawn prefix with the same AS path | | | and same attributes | | AW | Withdrawal of an announced prefix | | WW | Withdrawal of a withdrawn prefix (possible BGP error or a | | | data collection error) | +------+------------------------------------------------------------+ The following table classifies all the updates according to this taxonomy. +------+---------+ | Code | Count | +------+---------+ | AA+ | 607,093 | | AA- | 555,609 | | AA0 | 594,029 | | AA* | 782,404 | | AA | 195,707 | | WA+ | 238,141 | | WA- | 190,328 | | WA0 | 51,780 | | WA* | 30,797 | | WA | 77,440 | | AW | 627,538 | | WW | 0 | +------+---------+ Li & Huston Expires December 15, 2007 [Page 15] Internet-Draft BGP Stability Improvements June 2007 The interesting numbers here are those associated with BGP path hunting following a withdrawal, which are likely to be associated with the 607,093 AA+ updates and the 627,538 AW updates. But the population of these update types alone is not enough on its own to justify a conclusion that over 1.2 million updates are associated with path hunting as a precursor to prefix withdrawal events. The other salient factor that needs to be examined is the time distribution of updates, as the path hunting condition is associated with a rapid burst of updates. In looking at the time distribution of updates for the same prefix, there are some prominent peaks. The operation of the 30 second MRAI timer appears to be very prominent, and 934,391 updates occurred precisely 30 seconds after the previous update for the same prefix, and a total 1,636,093 updates for the same prefix occurred within 31 seconds of the previous update. That's the equivalent to 39% of the entire BGP update activity for the month. There are further local peaks at 30 second intervals at 60, 90 and all the way through to 240 seconds. Almost half of the BGP update activity occurs in a closely time-coupled manner. There are also smaller local peaks at 30 minute and 1 hour intervals. It is likely that these correspond to Route Flap Damping outcomes, where the damping period is typically one of these two values. Interestingly, there are also local peaks at 1, 2 and 3 day intervals. This is unlikely to be an artifact of Route Flap Damping, and is more likely to be the outcome of some form of time-managed traffic system that performs routing changes on a regular 24 hour basis. Another way of looking at this time distribution of updates is to construct update "sequences" where a pair of updates is considered to be part of the same sequence if it refers to the same prefix and is received within 65 seconds (or slightly longer than two MRAI Timer intervals) of any previous update for the same prefix. Only 28% of the updates for the month are not part of any sequence, while 26% of updates are part of a coupled update pair, and 46% of updates are part of sequences of 3 or more updates. Interestingly enough, changing the timer as to what defines a sequence does not alter the profile greatly. Extending the sequence timer to 125 seconds (or four MRAI Timer intervals) produces the outcome that 54% of updates are part of sequences of 3 or more updates, while reducing the sequence timer to 35 seconds produces the outcome that 36% of all updates are part of sequences of 3 or more updates. The approaches to flap damping to date have tended to look at flap damping as a persistent condition that lasts for hours or longer, and are an outcome of a strongly persistent announcement and withdrawal Li & Huston Expires December 15, 2007 [Page 16] Internet-Draft BGP Stability Improvements June 2007 characteristics that are assumed to be associated with some form of cyclical behavior of an underlying circuit. This now appears to be well wide of the mark in terms of capturing the profile of what appears to be redundant BGP updates that reflect transitory routing states that are not in any converged state. The question this prompts is whether there is any value in looking at BGP update patterns in the micro view rather than the macro? Can we identify, on the fly, update sequences that are highly likely to correlate to the BGP behavior of path hunting to a withdrawal and damp the intermediate path hunting states and simply propagate the resultant converged state? This was part of the intent of the MRAI timer, but rather than simply apply a uniform damping interval to update propagation, can we devise a selective algorithm that attempt to pinpoint routing transitions that are strongly associated with BGP path hunting? There are a number of observations here that appear to point to some value in considering this approach: o A BGP update generator may perform "update compression" by removing an already queued update when a further update that refers to the same prefix is generated. Thus, when using the MRAI timer, only the most recent state for each prefix is passed to the BGP peers, and any intermediate state that occurred during the MRAI-imposed quiescent time is suppressed. o Convergence in BGP appears to take longer than a single MRAI timer interval. As noted in the sequencing profile for updates, some 36% of all sequences take more than two MRAI timer intervals, or more than 60 seconds, to complete. o Path hunting in BGP is commonly represented as an update sequence of the form {AA+}* AW, i.e. a sequence of lengthening AS path lengths followed by a withdrawal. o Suppression of updates that lengthen the AS path length of a prefix does not implicitly create any risks of routing loop formation during the suppression period. If the peer had already selected a different path as the best path, then the update to a longer path would have no impact on the previous selection. If the peer was using the path advertised by this BGP speaker as its best path, then the suppression may cause the peer to continue to use this out-of-date path, but would not cause a path loop, as if the peer was listed on the longer path then the peer would already have a shorter path, and this update would not alter the peer's forwarding state. Li & Huston Expires December 15, 2007 [Page 17] Internet-Draft BGP Stability Improvements June 2007 So the profile of update sequences that appear to be effective targets for some form of local suppression are those that lengthen the AS path, and possibly also those updates that do not change the AS path Length, and are also part of a sequence of time-clustered updates for the same prefix. One approach to path length damping is to delay the processing an update if the update represents a lengthening of the AS path for an already announced prefix, selecting the AA+ updates. Furthermore, the updates that are of interest here are those that occur during BGP path hunting, so the length of time of the suppression should not be minutes or hours, but slightly over one MRAI time interval, or 35 seconds. If no further updates for this prefix are generated in this suppression interval, then the update is processed at the end of the suppression time, otherwise the suppressed update is replaced by its successor update. How effective would this form of path length damping be in the context of the BGP Update data set we've been examining here? The algorithm used to implement this damping response is to suppress the processing all AA+ updates by up to 35 seconds. If a further update for this prefix occurs during this suppression interval, then the suppressed update is ignored and the successor update is processed in stead. If this update represents a further lengthening of the AS path then it, too, is suppressed for 35 seconds. There are 607,093 AA+ updates in this set of suppressed updates, or some 15% of the total update load for the month. Path length damping would result in not processing 371,943 updates, or some 9.5% of the total update load. This result also indicates that 61% of all AA+ updates are followed by a subsequent update for the same prefix within one MRAI time interval. Decreasing the sensitivity of the suppression parameters to a little over 2 MRAI intervals, or 65 seconds, increases the number of unprocessed updates to 418,805, or an additional 1%, so it appears that a damping sensitivity of a single MRAI interval represents a suitable point of compromise between maximizing the number of damped BGP events without making the BGP convergence process significantly slower. Of these damped updates, how many are actually path hunt updates? Some 96,135 of these damped updates are immediately followed by a withdrawal within the 35 second suppression period, and a further 36,691 damped updates are followed by another suppressed AA+ update. This approach could be extended in a number of directions. One approach is to regard any update that does not reduce the AS path Li & Huston Expires December 15, 2007 [Page 18] Internet-Draft BGP Stability Improvements June 2007 length or withdraw the prefix as being a candidate for damping. In this case some 845,290 updates would be damped or 21% of all updates. Of these update just under one quarter, or 208,007 of these damped updates are followed by a withdrawal within one MRAI interval, and a further 474,234 of these damped updates are followed by an update with an AS path of the same or greater length. The implication being that path length damping removes around one fifth of the total BGP update volume without reducing the time to convergence for route withdrawal, nor the time for propagation of more preferred routing paths. Using this latter form of path length damping, over the month the average prefix update rate per second falls from 1.60 prefix updates per second to 1.22 prefix updates per second, with 0.38 damped updates per second on average. The average peak update rate per hour falls from 355 to 290 prefix updates per second using path length damping, an average reduction of 65 updates per second on the hourly peaks. 4.4. Prototyping, Testing and Pilot Deployment After some analysis, it would then be helpful to actually implement the most useful possible solutions in a number of BGP implementations. Since this is a change to BGP, extensive testing is going to be necessary and a period of pilot deployment will be required. Implementers, testers, and operators could help accelerate this portion of the project. 5. Acknowledgments This document builds on the work of [RFC2439] and we would like to thank Curtis Villamizar, Ravi Chandra, and Ramesh Govindan for their excellent work. We would like to thank Brian Carpenter and Robin Whittle for their helpful comments. 6. IANA Considerations This memo includes no requests to IANA. 7. Security Considerations This document raises no new security issues. Li & Huston Expires December 15, 2007 [Page 19] Internet-Draft BGP Stability Improvements June 2007 8. References 8.1. Normative References [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. [RFC4271] Rekhter, Y., Li, T., and S. Hares, "A Border Gateway Protocol 4 (BGP-4)", RFC 4271, January 2006. 8.2. Informative References [CN2000] Varadhan, K., Govindan, R., and D. Estrin, "Persistent Route Oscillations in Inter-domain Routing", Computer Networks vol. 32, pp. 1-16, 2000, . [Harmful] Bush, R., Griffin, T., and Z. Mao, "Route flap damping: harmful?", . [I-D.iab-raws-report] Meyers, D., "Report from the IAB Workshop on Routing and Addressing", draft-iab-raws-report-02 (work in progress), April 2007. [Potaroo0607] Huston, G., "Damping BGP", . [RFC2439] Villamizar, C., Chandra, R., and R. Govindan, "BGP Route Flap Damping", RFC 2439, November 1998. [RFC3345] McPherson, D., Gill, V., Walton, D., and A. Retana, "Border Gateway Protocol (BGP) Persistent Route Oscillation Condition", RFC 3345, August 2002. [RIPE-378] Smith, P. and C. Panigl, "RIPE Routing Working Group Recommendations on Route-flap Damping", . 8.3. Potential References [AggrWithdrawl] Raszuk, R., Patel, K., Appanna, C., and D. Ward, "BGP Aggregate Withdraw", draft-raszuk-aggr-withdraw-00 (work in progress), . Li & Huston Expires December 15, 2007 [Page 20] Internet-Draft BGP Stability Improvements June 2007 [FTCS-1999] Labovitz, C., Ahuja, A., and F. Jahanian, "Experimental Study of Internet Stability and Wide-Area Network Failures", FTCS 1999. [Infocom-1999] Labovitz, C., Malan, G., and F. Jahanian, "Origins of Internet Routing Instability", Infocom 1999. [Infocom-2001] Labovitz, C., Ahuja, A., Wattenhofer, R., and S. Venkatachary, "The Impact of Internet Policy and Topology on Delayed Routing Convergence", Infocom 2001. [Infocom-2005] Chandrashekar, J., Duan, Z., Zhang, Z., and J. Krasky, "Limiting path exploration in BGP", Infocom 2005. [PCC-2004] Duan, Z., Chandrashekar, J., Krasky, J., Xu, K., and Z. Zhang, "Damping BGP Route Flaps", IEEE International Conference on Performance, Computing, and Communications 2002. [R-BGP] Kushman, N., Kandula, S., Katabi, D., and B. Maggs, "R-BGP: Staying Connected In a Connected World", 4th USENIX Symposium on Networked Systems Design & Implementation 2007, . [Sigcomm-2000] Labovitz, C., Ahuja, A., Bose, A., and F. Jahanian, "Delayed Internet Routing Convergence", Sigcomm 2000. [Sigcomm-2002] Mao, Z., Govindan, R., Varghese, G., and R. Katz, "Route Flap Damping Exacerbates Internet Routing Convergence", Sigcomm 2002. [TON-1998] Labovitz, C., Malan, G., and F. Jahanian, "Internet Routing Instability", TON 1998. Li & Huston Expires December 15, 2007 [Page 21] Internet-Draft BGP Stability Improvements June 2007 Authors' Addresses Tony Li Cisco Systems, Inc. 170 W. Tasman Dr. San Jose, CA 95134 US Phone: +1 408 853 1494 Email: tli@cisco.com Geoff Huston Asia Pacific Network Information Centre Email: gih@apnic.net URI: http://www.apnic.net Li & Huston Expires December 15, 2007 [Page 22] Internet-Draft BGP Stability Improvements June 2007 Full Copyright Statement Copyright (C) The IETF Trust (2007). This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights. This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Intellectual Property The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79. Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr. The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf-ipr@ietf.org. Acknowledgment Funding for the RFC Editor function is provided by the IETF Administrative Support Activity (IASA). Li & Huston Expires December 15, 2007 [Page 23]