



RTGWG                                                 A. Abdelsalam, Ed.
Internet-Draft                                               C. Filsfils
Intended status: Informational                       Cisco Systems, Inc.
Expires: 3 September 2026                                        E. Ruan
                                                                  D. Cai
                                                           Alibaba Cloud
                                                            2 March 2026


                 IP Fast Reroute for BGP-Only Networks
            draft-abdelsalam-rtgwg-ipfrr-bgp-only-network-00

Abstract

   IP Fast Reroute (IPFRR) [RFC5714] is widely deployed in IP networks
   to provide protection against failures by invoking locally determined
   repair paths.  Traditional IPFRR deployments require the use of a
   link-state IGP to provide locally computed repair paths.  This
   document provides the problem statement and outlines a solution to
   enable IPFRR computation and deployment in BGP-only networks.  The
   solution maintains standard BGP routing operations while providing
   fast reroute capabilities comparable to IGP-based deployments.

Status of This Memo

   This Internet-Draft is submitted in full conformance with the
   provisions of BCP 78 and BCP 79.

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF).  Note that other groups may also distribute
   working documents as Internet-Drafts.  The list of current Internet-
   Drafts is at https://datatracker.ietf.org/drafts/current/.

   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."

   This Internet-Draft will expire on 3 September 2026.

Copyright Notice

   Copyright (c) 2026 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents (https://trustee.ietf.org/
   license-info) in effect on the date of publication of this document.



Abdelsalam, et al.      Expires 3 September 2026                [Page 1]

Internet-Draft         IPFRR in BGP-Only Networks             March 2026


   Please review these documents carefully, as they describe your rights
   and restrictions with respect to this document.  Code Components
   extracted from this document must include Revised BSD License text as
   described in Section 4.e of the Trust Legal Provisions and are
   provided without warranty as described in the Revised BSD License.

Table of Contents

   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   2
   2.  Problem Statement . . . . . . . . . . . . . . . . . . . . . .   2
   3.  IPFRR for BGP-only networks . . . . . . . . . . . . . . . . .   3
     3.1.  Topology Collection . . . . . . . . . . . . . . . . . . .   3
     3.2.  The IPFRR Agent . . . . . . . . . . . . . . . . . . . . .   4
   4.  Considerations for BGP Policies . . . . . . . . . . . . . . .   5
   5.  Security Considerations . . . . . . . . . . . . . . . . . . .   5
   6.  References  . . . . . . . . . . . . . . . . . . . . . . . . .   5
     6.1.  Normative References  . . . . . . . . . . . . . . . . . .   5
     6.2.  Informative References  . . . . . . . . . . . . . . . . .   5
   Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .   6

1.  Introduction

   IP Fast Reroute (IPFRR) [RFC5714] is widely deployed in IP networks
   to provide protection against failures by invoking locally determined
   repair paths.

   In Data Center (DC) networks with extensive Equal-Cost Multi-Path
   (ECMP), operators often rely on local fast ECMP re-hashing to protect
   against failures of ECMP members.  However, some failures are not
   addressed by this ECMP re-hashing and require additional fast reroute
   solutions for comprehensive protection.  The requirements and
   mechanisms for IPFRR in Artificial Intelligence (AI) DC networks and
   DCI environments are described in [I-D.draft-clad-rtgwg-ipfrr-aiml].

   Deploying IPFRR in networks traditionally requires running a link-
   state IGP.  This document describes an FRR solution for BGP-only
   networks where no IGP is present, like Massively Scalable Data
   Centers (MSDCs), AI DCs, and DCI networks.

2.  Problem Statement

   [RFC5714] specifies the IPFRR framework assuming link-state IGP
   deployment.  This framework relies on two fundamental capabilities:

   1.  Complete topology visibility through Link-State Databases (LSDBs)

   2.  Infrastructure to compute paths using the LSDB and install repair
       paths in the forwarding table



Abdelsalam, et al.      Expires 3 September 2026                [Page 2]

Internet-Draft         IPFRR in BGP-Only Networks             March 2026


   BGP-only networks face the following challenges in deploying IPFRR:

   *  Lack of Topology Information: As a path-vector protocol, BGP
      advertises reachability information without topology details.  BGP
      routers maintain no equivalent to an IGP LSDB.

   *  No Native Topology-Based Computation: BGP's best-path selection
      process [RFC4271] operates on path attributes, not topology.  BGP
      has no inherent mechanism to compute paths based on network
      topology.

   *  Policy Complexity: Unlike IGPs where forwarding follows shortest
      paths, BGP policies at intermediate hops can alter forwarding
      decisions, complicating repair path computation.  See Section 4
      for more details.

   These limitations prevent direct application of IPFRR techniques in
   BGP-only environments.

3.  IPFRR for BGP-only networks

   This document describes a solution that enables IPFRR computation and
   installation of repair paths in BGP-only networks by leveraging
   topology information obtained via BGP-LS.

   The solution does not modify base BGP routing protocol operations in
   the BGP-only network fabric, which continues to provide routing using
   the BGP best path selection process [RFC4271].

3.1.  Topology Collection

   [I-D.ietf-idr-bgp-ls-bgp-only-fabric] defines a solution for
   obtaining a detailed topology view, similar to that available when
   using link-state routing protocols, in networks where BGP is used as
   the only routing protocol.  It specifies extensions to the BGP Link-
   State (BGP-LS) address family and procedures for advertising topology
   information in BGP-only networks.

   With this topology information, IPFRR computations can be performed
   on the BGP-only network, using the same principles as in link-state
   IGPs (OSPF, IS-IS).  The primary difference lies in how the topology
   database is populated.

   The extensions in [I-D.ietf-idr-bgp-ls-bgp-only-fabric] have been
   implemented and available as open-source in the FRR Routing Stack
   [FRRouting].





Abdelsalam, et al.      Expires 3 September 2026                [Page 3]

Internet-Draft         IPFRR in BGP-Only Networks             March 2026


3.2.  The IPFRR Agent

   The solution introduces an independent logical entity, referred to as
   the "IPFRR Agent" which is responsible for IPFRR computation and
   programming of repair paths on behalf of BGP routers that have
   registered to it.  Depending on the deployment model, an IPFRR Agent
   may service one or multiple BGP routers.

   The IPFRR Agent is an application that can be located within a
   network node, on an out-of-network server, within a controller, or
   elsewhere.

   The IPFRR Agent continuously collects the real time network topology
   information from a centralized BGP-LS speaker.  On behalf of each of
   the BGP router that has registered to it, the agent runs the IPFRR
   computations, and programs the resulting repair paths on the
   corresponding BGP router.

   The workflow to deploy IPFRR in a BGP-only network via the IPFRR
   Agent involves the following steps:

   1.  Topology collection: All BGP routers report their local topology
       view to one or multiple centralized BGP-LS speakers.  BGP Route
       Reflectors (RRs) are typically used here to scale the
       distribution of this information.

   2.  Topology consolidation: The centralized BGP-LS speakers
       consolidate the BGP-LS topology information received from all BGP
       routers to build a database representing the full network
       topology.

   3.  Topology retrieval: The IPFRR Agent retrieves the full real time
       network topology view from the centralized BGP-LS speaker.  The
       IPFRR Agent updates its topology database upon receiving BGP-LS
       advertisements.  The IPFRR Agent may be co-located with or
       external to the centralized BGP-LS speaker.

   4.  IPFRR computation: The IPFRR Agent computes the IPFRR repair
       paths for each of its registered BGP routers, using its topology
       database.

   5.  Programming: The IPFRR Agent installs the computed IPFRR repair
       paths on the corresponding registered BGP router using a standard
       or well-known southbound API (e.g., NETCONF, RESTCONF, gRPC, or
       local APIs).

   The details of the IPFRR computation algorithm are outside the scope
   of this document.



Abdelsalam, et al.      Expires 3 September 2026                [Page 4]

Internet-Draft         IPFRR in BGP-Only Networks             March 2026


4.  Considerations for BGP Policies

   In link-state IGP networks, IPFRR computation relies solely on the
   network topology, as forwarding decisions are determined by the
   shortest path algorithm without local policy modifications.  However,
   in eBGP hop-by-hop environments, BGP policies applied on intermediate
   nodes may affect forwarding decisions, and therefore IPFRR
   computation outcomes.  These policies MUST be accounted for during
   IPFRR computation to ensure repair paths are viable and loop-free.

5.  Security Considerations

   The solution relies on BGP-LS for topology population.  Standard BGP-
   LS security considerations apply.  Additional care must be taken to
   secure the interface between the IPFRR Agent and BGP speakers or
   controllers, especially when using external southbound APIs.

6.  References

6.1.  Normative References

   [I-D.ietf-idr-bgp-ls-bgp-only-fabric]
              Talaulikar, K., MahendraBabu, A. B., Filsfils, C.,
              Ananthamurthy, K., Zandi, S., Dawra, G., and M. Durrani,
              "BGP Link-State Extensions for BGP-only Networks", Work in
              Progress, Internet-Draft, draft-ietf-idr-bgp-ls-bgp-only-
              fabric-04, 20 October 2025,
              <https://datatracker.ietf.org/doc/html/draft-ietf-idr-bgp-
              ls-bgp-only-fabric-04>.

   [RFC4271]  Rekhter, Y., Ed., Li, T., Ed., and S. Hares, Ed., "A
              Border Gateway Protocol 4 (BGP-4)", RFC 4271,
              DOI 10.17487/RFC4271, January 2006,
              <https://www.rfc-editor.org/info/rfc4271>.

   [RFC5714]  Shand, M. and S. Bryant, "IP Fast Reroute Framework",
              RFC 5714, DOI 10.17487/RFC5714, January 2010,
              <https://www.rfc-editor.org/info/rfc5714>.

6.2.  Informative References

   [FRRouting]
              "Add support for BGP-LS for BGP fabric", 2026,
              <https://github.com/FRRouting/frr/pull/20726>.







Abdelsalam, et al.      Expires 3 September 2026                [Page 5]

Internet-Draft         IPFRR in BGP-Only Networks             March 2026


   [I-D.draft-clad-rtgwg-ipfrr-aiml]
              "IP Fast Reroute for AI/ML Fabrics", 2026,
              <https://datatracker.ietf.org/doc/draft-clad-rtgwg-ipfrr-
              aiml>.

Authors' Addresses

   Ahmed Abdelsalam (editor)
   Cisco Systems, Inc.
   Italy
   Email: ahabdels@cisco.com


   Clarence Filsfils
   Cisco Systems, Inc.
   Belgium
   Email: cf@cisco.com


   Eddie Ruan
   Alibaba Cloud
   United States of America
   Email: eddie.ruan@alibaba-inc.com


   Dennis Cai
   Alibaba Cloud
   United States of America
   Email: d.cai@alibaba-inc.com






















Abdelsalam, et al.      Expires 3 September 2026                [Page 6]
