Internet Draft August 2006 Network Working Group Manav Bhatia Internet Draft Lucent Technologies Joel M. Halpern Paul Jakma Expires: January 2007 Sun Microsystems Advertising Multiple NextHop Routes in BGP draft-bhatia-bgp-multiple-next-hops-01.txt 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 August 2006 Copyright Notice Copyright (C) The Internet Society (2006). Abstract This document describes an extensible mechanism that allows a BGP speaker to advertise multiple BGP paths for a destination to its peers, by describing a new BGP capability, termed "Multiple-Hop Capability". The mechanisms described in this document are applicable to all routers, both those with the ability to inject multiple routing entries in their forwarding table and those without. Bhatia, Halpern and Jakma [Page 1] Internet Draft August 2006 Conventions used in this document 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 [KEYWORDS] Table of Contents 1. Introduction...................................................2 2. Multiple-Hop Capability........................................3 2.1 Multiple-Hop attribute - MULTIPLE_HOP......................5 3. Operation when both peers are Multiple-Hop capable.............6 3.1 Advertisement of Multiple-Hop BGP routes...................7 3.2 Withdrawal Procedures......................................7 3.3 Procedures for the Receiving Speaker.......................8 3.4 Working with Multiple-Hop capable IBGP peers...............8 3.5 Implicit Withdrawal for one of the Next-Hops...............9 4. Multiprotocol Extensions to BGP................................9 5. Security Considerations.......................................10 6. Acknowledgements..............................................10 7. IANA Considerations...........................................10 8. References....................................................10 8.1 Normative References......................................10 8.2 Informative References....................................11 9. Appendix A....................................................11 9.1 Suboptimal Routing in Route Reflector clients.............11 9.2 Avoiding Persistent Route Oscillations....................12 9.3 eBGP mesh scaling at IXes via Route Servers...............15 9.4 Advertising a subset of routes in BGP.....................15 9.5 Equal Cost Multiple Path BGP..............................16 10. Author’s Address.............................................16 11. Intellectual Property Statement..............................17 1. Introduction Currently BGP [BGP4] speakers cannot announce multiple paths, even if it is desirable in certain scenarios. This is because the BGP specification allows only one "best" route to be inserted into the Loc-RIB, and to be announced to other BGP speakers. If another route for a destination that has previously been announced to a BGP peer, is sent later, then the receiver “implicitly withdraws” the former route and replaces it with the new one. Because of this behavior, BGP speakers are never able to advertise multiple paths for the same destination to their peers. Lifting this restriction would have benefit for at least the following scenarios in BGP: Bhatia, Halpern and Jakma [Page 2] Internet Draft August 2006 o Persistent route-oscillation conditions in BGP [MED] o eBGP mesh scaling at Internet Exchanges o Interaction between ECMP capable BGP speakers The first concerns route-reflectors [RR], where in certain topologies, persistent route-oscillation conditions can arise due to the clients of route-reflectors being never fully informed of each others best paths, particularly where MED/Router ID values are considered as part of the best-path selection. If BGP were to provide a means to allow route-reflectors to share all the collective best-paths with its clients, then these conditions could be alleviated, as has been shown in the Appendix. The second concerns scaling of eBGP meshes at Internet Exchanges (referred to as an IX from now on, or IXes in the plural). IX operators have deployed eBGP route-servers, in a variety of guises, in order to reduce the need for customers to establish direct sessions with other customers. These route-servers however have severe limitations because of the single-path restriction in BGP. Removing this limitation would allow for efficient deployment of IX route-servers. The third concerns BGP implementations which are capable of considering multiple routes for inclusion into their RIB, and hence likely their FIB, but do not have a way to relay the full resulting state of their BGP RIB to their peers. This document specifies the mechanism by which Multiple-Hop operates; however it will not attempt to fully describe the usages. In particular this document anticipates that the ECMP scenario will be described fully in another document, as it would have to be even if documented without consideration of the Multiple-Hop capability. It is anticipated however that any speaker implementing the functionality described in this document would be able to interoperate with Multiple-Hop capable route-servers and route- reflectors, just as BGP speakers interoperate with Route-Reflectors in the absence of the Multiple-Hop capability. 2. Multiple-Hop Capability Multiple Hop capability is a new capability that can be used by a BGP speaker to indicate its ability to understand Multiple-Hop Updates from a remote peer. This capability is defined as follows: Bhatia, Halpern and Jakma [Page 3] Internet Draft August 2006 Capability Code: TBD Capability Length: Variable Capability Values: Consists of one or more of the tuples as follows: +--------------------------------------------------+ | Address Family Identifier (16 bits) | +--------------------------------------------------+ | Subsequent Address Family Identifier (8 bits) | +--------------------------------------------------+ | Flags for the Address Family (8 bits) | +--------------------------------------------------+ Figure 1 The use and meaning of the fields are as follows: Address Family Identifier This field carries the identity of the Network Layer protocol for which the Multiple Hop support is advertised. Presently defined values for this field are specified in [IANA-AFI]. Subsequent Address Family Identifier (SAFI): This field provides additional information about the type of the Network Layer Reachability Information carried in the attribute. Presently defined values for this field are specified in [IANA-SAFI]. Flags for Address Family: This field contains bit flags for the . 0 1 2 3 4 5 6 7 +-+-+-+-+-+-+-+--+ |R|R|R|R|R|R|R|RM| +-+-+-+-+-+-+-+--+ R Reserved: MUST be set to zero by the sender and ignored by the receiver. RM Receive Multiple Indicates that the speaker is interested in receiving additional Bhatia, Halpern and Jakma [Page 4] Internet Draft August 2006 BGP paths, other than just the best path from the receiver. A speaker sets this bit in its MULTIPLE_NEXT_HOP capability to indicate that it is prepared to receive additional path advertisements, beyond just the best path, by way of the MULTIPLE_NEXT_HOP capability. As such, speakers implementing the MULTIPLE_NEXT_HOP capability MUST not send additional paths, beyond the single best path allowed by BGP-4 [BGP4], unless the remote speaker has indicated its preparedness with the RM bit. 2.1 Multiple-Hop attribute - MULTIPLE_HOP This attribute is an optional, non-transitive attribute that can be used for advertising multiple next-hops associated with a NLRI. The attribute data contains one or more tuples of (AFI,SAFI, List of Next Hop Information), where each tuple is encoded as shown below: +------------------------------------------------+ | Address Family Identifier (2 octets) | +------------------------------------------------+ | Subsequent Address Family Identifier (1 octet) | +------------------------------------------------+ | Number of Next Hops (1 octet) | +------------------------------------------------+ | Length of the First Next Hop (1 octet) | +------------------------------------------------+ | Network Address of First Next Hop (variable) | +------------------------------------------------+ | Length of the Second Next Hop (1 octet) | +------------------------------------------------+ | Network Address of Second Next Hop (variable) | +------------------------------------------------+ | . . . | | . . . | +------------------------------------------------+ | Length of the Nth Next Hop (1 octet) | +------------------------------------------------+ | Network Address of Nth Next Hop (variable) | +------------------------------------------------+ Figure 2 The various fields are defined as follows: Bhatia, Halpern and Jakma [Page 5] Internet Draft August 2006 Address Family Identifier: The AFI field carries the identity of the Network Layer protocol associated with the Network Address that follows. Subsequent Address Family Identifier: The SAFI field in combination with the Address Family Identifier field identifies the Network Layer context associated with the Network Address of the Next Hop(s). Number of Next-Hops: This field carries the total number of Multiple- Hop BGP routes for the given NLRI. Length of Nth Next Hop Network Address: A 1 octet field whose value expresses the length of the "Network Address of Next Hop" field as measured in octets. For IPv6 routes the value shall be set to 16, when only a global address is present, or 32 if a link-local address is also included in the Next Hop field [BGP-IPv6]. Network Address of Nth Next Hop: This is a variable length field that contains the Network Address of the next router on the path to the destination. The N next-hops listed in the MULTIPLE_HOP path attribute define the Network Layer address of the routers that should be used as next-hops to the destinations listed in the UPDATE message. 3. Operation when both peers are Multiple-Hop capable In the following sections, "Local speaker" refers to a router which is advertising the BGP Multiple-Hop routes, and the "Receiving Speaker" refers to a router that peers with the former to accept multiple BGP routes for a destination. Consider that the Multiple-Hop Capability has been exchanged between the Local speaker and the Receiving speaker, and a BGP session between them is established. The following sections detail the procedures that shall be followed by the Local speaker as well as the Receiving speaker once the Multiple-Hop capability has been exchanged, and the local speaker wants to advertise some BGP Multiple-Hop routes. Note that for operation within the confines of this document and BGP, the local speaker almost certainly will be acting as an eBGP route- server or iBGP route-reflector, with the receiver asserting the RM bit in the Multiple-Hop capability, and therefore acting as a client of that speaker. Other uses, such as ECMP speakers exchanging Multiple-Hop routes will require further consideration, not addressed in this document as Bhatia, Halpern and Jakma [Page 6] Internet Draft August 2006 stated previously, considerations not per se related to the Multiple- Hop capability itself. 3.1 Advertisement of Multiple-Hop BGP routes The extensions proposed in this draft allow BGP paths to be identified by their NLRI and next-hop address, rather than just by their NLRI. This extended identification is indicated by the presence of the MULTIPLE_HOP attribute. Given that this is used when there are multiple paths sharing NLRI, this attribute allows for the representation of multiple such paths in a single advertisement. Thus between Multiple-Hop capable speakers, the MULTIPLE_HOP attribute MUST be used in addition to the existing NEXT_HOP in order to announce multiple next-hops for the destinations listed in the NLRI field of the UPDATE message. All prefixes announced using this attribute MUST NOT replace the previous advertisements and thus, multiple BGP paths for a prefix can be advertised by the Local Speaker. If the same prefix is later announced with ONLY the NEXT_HOP attribute then it MUST be taken as an implicit withdraw for all the previous paths advertised by that peer for that destination. It should be noted that transmission of multiple paths is only valid for the same NLRI that differ on the next-hop. An UPDATE message which contains feasible routes and carries MULTIPLE_HOP and no NEXT_HOP attribute MUST NOT be considered as an implicit withdrawal. The Receiving Speaker MUST append these routes in its Adj-RIBs-In [BGP4], as additional paths to that destination. When advertising multiple paths which do not have identical path attributes, separate BGP UPDATE messages MUST be sent, each with a MULTIPLE_HOP attribute even if there is only one next-hop in each MULTIPLE_HOP attribute. Presence of MULTIPLE_HOP suppresses route replacement at the receiving end. 3.2 Withdrawal Procedures An UPDATE message which contains an IP address prefix in the WITHDRAWN ROUTES marks all the associated routes as being no longer available for use. An UPDATE message consisting of an IP address prefix in the NLRI field and only the NEXT_HOP attribute implicitly withdraws all the routes to that address prefix and replaces it with the one advertised by the NEXT_HOP. Bhatia, Halpern and Jakma [Page 7] Internet Draft August 2006 An UPDATE message which contains an IP address prefix in the WITHDRAWN ROUTES and the MULTIPLE_HOP attribute only removes the path associated with that next-hop. An UPDATE message announced with a MULTIPLE_HOP attribute for a given IP address prefix implicitly withdraws any previous route announced with the same next-hop. 3.3 Procedures for the Receiving Speaker The Receiving Speaker upon receiving the MULTIPLE_HOP attribute will understand that the Local Speaker has advertised Multiple-Hop BGP routes. Within a single UPDATE message all the prefixes will have identical attributes, except for the next-hops, which will be carried in the MULTIPLE_HOP attribute. A series of further UPDATE messages for the same NLRI, with or without the same set of attributes and containing the MULTIPLE_HOP attribute will be understood to be additive. Each UPDATE message would append these additional feasible routes, to the appropriate Adj-RIBs-In, where after the receiving speaker may run its normal decision process to select the best path to install in its Local-RIB. Upon receiving an UPDATE message for the same NLRI, without the MULTIPLE_HOP attribute, the receiver will consider this as a replacement route for all the previously announced routes to that destination. If the BGP Speaker wants to withdraw all the BGP routes for a particular address prefix then it can send a normal BGP UPDATE message listing the IP address prefix in the WITHDRAWN ROUTES field. The Receiving Speaker upon receiving this message MUST remove all the routes associated with that destination. If the Receiving Speaker receives an UPDATE message with the MULTIPLE_HOP attribute listing both, the feasible and the unfeasible routes, then it MUST consider the path attributes for the feasible routes. All the destinations listed in the WITHDRAWN ROUTES MUST be removed as per [BGP4]. 3.4 Working with Multiple-Hop capable IBGP peers This section explains how multiple-hop feature will work in the normal scenarios. Assume that the two IBGP speakers A and B exchange this capability. Consider a case where A receives multiple UPDATE messages for NLRI X with next-hops Nj, Nk and Nm. Assume that all these routes are valid Bhatia, Halpern and Jakma [Page 8] Internet Draft August 2006 and A wants to pass on this set to B. Also assume that Nj and Nk share the same path attributes (Origin, AS Path, Local Pref, etc) and can be thus advertised in a single UPDATE message. A makes an UPDATE message and uses the MULTIPLE_HOP path attribute. It puts the AFI, SAFI, number of next-hops as 2, length of the first next-hop Nj, network address of Nj, length of Nk and the network address of Nk. When this UPDATE message reaches B, it looks at the MULTIPLE_HOP attribute and understands that there are multiple routes to reach X. It inserts the two routes for X with the next-hops Nj and Nk in its Adj-RIBs-In. A also needs to announce the remaining route to X with next-hop Nl. It makes an UPDATE message, fills the path attributes, and uses the MULTIPLE_HOP attribute to encode next-hop information about Nl. This UPDATE message is sent to B. When B receives this UPDATE message it knows that this is not a replacement route for X as it comes with the MULTIPLE_HOP attribute. It simply appends this new route in its adj-RIBs-In, runs the decision process, and proceeds as normal. Assume that at some point later, A needs to withdraw the route associated with the tuple [X, nexthop Nk]. It makes an UPDATE message, puts X in the WITHDRAWN ROUTES and inserts the MULTIPLE_HOP attribute, encoding the next-hop Nk inside. When B receives this UPDATE message it understands that A wants to remove one (or more) of the routes associated with X. To determine which exact route(s) needs to be removed, it looks at the MULTIPLE_HOP attribute and goes about removing all the routes associated with the next-hops listed therein. 3.5 Implicit Withdrawal for one of the Next-Hops In the same scenario to replace a route associated with the tuple [X, next-hop Nk], A can advertise a fresh route with a new set of path attributes. B would consider the new advertisement as an implicit withdrawal for the previously announced route for the tuple [X, next- hop Nk]. 4. Multiprotocol Extensions to BGP Since the MULTIPLE_HOP includes both the AFI and SAFI, it is possible to advertise multiple MPBGP routes. In this case, MP_REACH_NLRI [MBGP] attribute shall carry the NLRI information and MULTIPLE_HOP the information about the additional next-hops. Bhatia, Halpern and Jakma [Page 9] Internet Draft August 2006 To suppress route replacement the additional routes must be advertised by keeping the length of the next-hop as 0 in the MP_REACH_NLRI attribute. The same should be encoded in the MULTIPLE_HOP attribute. 5. Security Considerations This extension to BGP does not change the underlying security issues inherent in the existing BGP. 6. Acknowledgements The authors would like to thank Tony Li, Arnold Nipper and Curtis Villamizar for their valuable comments and suggestions on the earlier versions of this draft from which the current work has been derived. 7. IANA Considerations IANA needs to assign a capability code to the Multiple Hop capability 8. References 8.1 Normative References [BGP-CAP] Chandra, R. and J. Scudder, "Capabilities Advertisement with BGP-4", RFC 3392, November 2002 [BGP4] Rekhter, Y., Li, T. and Hares, S., "A Border Gateway Protocol 4 (BGP-4)", RFC 4271, March 1995 [RR] Chandra, R., Bates, T., and E. Chen, "BGP Route Reflection - An Alternative to Full Mesh Internal BGP (IBGP)", RFC 4456, April 2006 [BGP-IPv6] Marques, P. and F. Dupont, "Use of BGP-4 Multiprotocol Extensions for IPv6 Inter-Domain Routing", RFC 2545, March 1999. [MBGP] Chandra, R., Rekhter, Y., Bates, T., and D. Katz, "Multiprotocol Extension for BGP-4", draft-ietf-idr-rfc2858bis-10.txt (work in progress) [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", RFC 2119, BCP 14, February 2001. [IANA_AFI] http://www.iana.org/assignments/address-family-numbers [IANA-SAFI]http://www.iana.org/assignments/safi-namespace Bhatia, Halpern and Jakma [Page 10] Internet Draft August 2006 8.2 Informative References [MED] Retana, A., Walton, D., McPherson, D., and V. Gill, "Border Gateway Protocol (BGP) Persistent Route Oscillation Condition", RFC 3345, August 2002. [COMM] Chandra, R., Trania, P. and Li, T.,”BGP Communities Attribute”, RFC 1997, August 1996 9. Appendix A This section explains some scenarios where advertising multiple BGP paths may prove to be useful. 9.1 Suboptimal Routing in Route Reflector clients Route Reflection can result in suboptimal routing due to the client not having full visibility to all the BGP paths in the AS. This is because the RR selects the best path and reflects only that best path to its clients. In case the RR has equal cost BGP routes, then it shall select the one based on the lower Router ID. As a result, the clients do not receive the full view of the available paths, or at least the paths that are equidistant from the RR. This can result in suboptimal routing from the client's perspective. A client may have selected a different best path if more paths had been made visible to it. With Multiple-hop BGP, the RR can advertise all the equal cost BGP routes that it has to its client, giving the client more options to choose from. The extensions proposed in this draft provide provision for the RR to reflect all the routes to its clients. Bhatia, Halpern and Jakma [Page 11] Internet Draft August 2006 9.2 Avoiding Persistent Route Oscillations ---------------------------------- / AS X \ | ----- | | / \ | | | | | | | RR | | | \ / | | -/+\- | | c1 / \ c2 | | ---- / \ ---- | | / \ / \ / \ | | ( Ra ) ( Rb ) | | \ / \ / | | -/\-- ------ | | / \ \ | | / \ \ | \ / \ \ / --/------\--------------------\---- / \ \ / --------------------------- / / \ --\-- \ --/- | \ / \ | // \\ | \ | | | | R2 | | \ | R3 | | | | | -\-- \ / | \\ // | / \ ----- | ---- | | | | AS Y | | R1 | | | \ / | | ---- | \ AS Z / ----------------------------- Figure 3 Consider the topology as shown in Figure 1. Say, AS X consists of Route Reflector (RR) and two clients Ra and Rb. Ra is connected to R2 in AS Y and R1 in AS Z. Rb is connected to R3 in AS Z. Assume that the Router ID of R1 < R2 and IGP cost c1 < c2. The dashed lines between the routers shows BGP peering. Assume that the BGP speakers in AS Y and AS Z receive a BGP UPDATE for 10.0.0.0/8 from AS W. Assume that they advertise the following path attributes to BGP speakers in AS X: R2: NLRI 10.0.0.0/8, AS_PATH Y W, MED 100, NEXT_HOP R2 Bhatia, Halpern and Jakma [Page 12] Internet Draft August 2006 R1: NLRI 10.0.0.0/8, AS_PATH Z W, MED 300, NEXT_HOP R1 R3: NLRI 10.0.0.0/8, AS_PATH Z W, MED 200, NEXT_HOP R3 Scenario 1: Traditional BGP in AS X The following events happen: 1. Ra receives UPDATE messages from R2 and R1. Since they are from different ASes, MEDs are not compared and the tie breaks on the lower Router ID. Since R1 < R2, route from R1 is selected and advertised to the RR. Ra thus has the following path as the best one for 10.0.0.0/8: AS_PATH Z W, MED 300, NEXT_HOP R1 2. Rb receives the UPDATE from R3, installs this and advertises the same to the RR. Rb thus has the following path for 10.0.0.0/8: AS_PATH Z W, MED 200, NEXT_HOP R3 3. RR receives two UPDATE messages from its clients. Since the neighboring AS is the same in both of them, the tie breaks on the route having the lower value of MED. It thus selects the route it learns from Rb as the best one and advertises this to Ra. 4. Ra now has all the three paths. Route learnt from Rb wins over the route learnt from R1 (lower MED) and the route learnt from R2 wins over the route learnt from Rb (EBGP > IBGP). 5. Ra thus sends an implicit WITHDRAW to the RR, replacing the earlier announcement with the route learnt from R2. 6. RR thus has the following paths for 10.0.0.0/8: AS_PATH Y W, MED 100, NEXT_HOP R2 AS_PATH Z W, MED 200, NEXT_HOP R3 It selects the first path because the IGP cost to reach the NEXT_HOP (R2) is lesser for the first one. It thus, advertises this path to Rb and sends a WITHDRAW message to Ra, removing the path it had initially announced (one learnt from Rb) 7. Ra receives the WITHDRAW message from the RR and removes the path. Nothing is done as it is currently not the best path. 8. Rb receives the advertisement from RR, but doesn't do anything, as the path learnt from R3 is better (EBGP > IBGP). Bhatia, Halpern and Jakma [Page 13] Internet Draft August 2006 9. Ra at this time has only two routes. One, learnt from R1 and the other learnt from R2: AS_PATH Z W, MED 300, NEXT_HOP R1 AS_PATH Y W, MED 100, NEXT_HOP R2 It has selected the route learnt from R2. After some time, this router runs its scanner process for validating the NEXT_HOPs. There it runs the best path algorithm and finds that the route learnt from R1 is better than the route learnt from R2, because of the lower Router ID. 10.Ra sends an implicit WITHDRAW to RR, replacing the earlier announcement with the route learnt from R2. 11... The loop follows and it cycles again and again. Scenario 2: Multiple-Hop BGP is implemented in AS X 1. If everything happens the same as in the preceding example then Ra will have two paths to reach 10.0.0.0/8. Since everything else is the same, it will advertise both these routes to the RR. Note that Ra will not look at the Router ID, etc. for tie breaking if Multiple-Hop capabilities are implemented. 2. RR will now have three paths for 10.0.0.0/8. Path 3, from Rb and Paths 1 and 2 from Ra. Path 1: AS_PATH Y W, MED 100, NEXT_HOP R2 Path 2: AS_PATH Z W, MED 300, NEXT_HOP R1 Path 3: AS_PATH Z W, MED 200, NEXT_HOP R3 Out of Path 2 and Path 3, it will select Path 3 (lower MED).From Path 1 and Path 3, it will select Path 1, based on the lower IGP cost. RR thus selects the Path 1 as the best route. 3. RR will advertise the new path to Rb. Rb will thus have the following two paths: Path 1: AS_PATH Y W, MED 100, NEXT_HOP R2 Path 2: AS_PATH Z W, MED 200, NEXT_HOP R3 Bhatia, Halpern and Jakma [Page 14] Internet Draft August 2006 Path 2 will win because of the EBGP > IBGP rule, and it will continue using R3. There is thus, no change on Rb and it continues using the same path as before. 4. The network is stable and there are no route oscillations. 9.3 eBGP mesh scaling at IXes via Route Servers IXes today sometimes offer their customers the facility to peer with a neutral IX route-server as a means to reduce the direct peering requirements for their customers. The peering overhead may be considerable given the many hundreds of ASes which may be present at some of the larger IXes today, and it is quite plausible that IXes will continue to grow in terms of attached customers and ASes. However, the single-path limitation of BGP imposes great operational difficulty in allowing such a route-server to be effective. There are typically two kinds of route-server, one which is a normal BGP speaker and simply provides a single-best-path-for-all service, and the type which are configured with each customer’s policies and calculate the best-path separately for each. Both approaches have their limitations: o Route-servers which simply advertise the current best known IX path according to normal BGP procedures, without applying any customer-specific policy, require the customers to often still establish direct sessions with each other for cases where they wish to apply policy. Much of the scaling benefits are never realised. o Route-servers which apply policy on their customers behalf, selecting the best-path on a per-customer basis and then advertising each customer a tailor-made best-path, require extensive co-ordination of policy between the IX operators and each of their customers. Further, it may be difficult for customers to keep their policies private due the operational requirements of policy co-ordination between IX and customer. If there were a mechanism in BGP to allow an IX route-server to pass all other advertisements to a customer peer, without performing any path selection or applying any policy, then this would remove the need for policy co-ordination between each customer and the IX, and address the other shortcomings listed above. Such a mechanism would be easy for both the IX operator and each customer to deploy and maintain. 9.4 Advertising a subset of routes in BGP Bhatia, Halpern and Jakma [Page 15] Internet Draft August 2006 Providers can tag some selected routes with certain communities [COMM]. An administrator could write a policy that would advertise all the paths carrying a known community within that AS to another router capable of understanding the Multiple-Hop extensions. This is a form of policy implementation and a detailed study of what could be achieved using such techniques is beyond the scope of this draft. 9.5 Equal Cost Multiple Path BGP Currently some implementations, when they receive multiple equal cost BGP routes from different peers, are able to insert all of them (or a subset of those, based on their local policies) in their forwarding table to locally split the load for the destination, while announcing only one "best" BGP path to its other peers. This however has implications for those other peers which receive such an announcement from this ECMP capable BGP speaker. The implication, as per route aggregation, is these other peers potentially will not posses the full path information, which can lead to loops. Hence, such an ECMP capable BGP speaker can only enable this feature if great care is taken, if at all, or must act as if it had aggregated the set of routes concerned. While this document does not directly address the question of ECMP, the mechanism introduced can be built upon in order to do so. It would be feasible to introduce additional semantics on top of the Multiple-Nexthop Capability so as to allow the ECMP BGP speaker to fully communicate the details of all the paths it is forwarding on, and hence allow those other peers to have full visibility of path information and be able to avoid selecting paths which would otherwise loop, while still maintaining compatibility with speakers not implementing ECMP and Multiple-Hop. 10. Author’s Address Manav Bhatia Lucent Technologies Email: manav@lucent.com Joel M. Halpern Email: joel@stevecrocker.com Paul Jakma Sun Microsystems Email: paul.jakma@sun.com Bhatia, Halpern and Jakma [Page 16] Internet Draft August 2006 11. Intellectual Property Statement 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. Disclaimer of Validity 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 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. Copyright Statement Copyright (C) The Internet Society (2006). 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. Acknowledgment Funding for the RFC Editor function is currently provided by the Internet Society. Bhatia, Halpern and Jakma [Page 17]