Network Working Group                                    Pierre Francois
Internet-Draft                          Universite catholique de Louvain
Intended status: Informational                            Bruno Decraene
Expires: April 29, 2010 28, 2011                                   France Telecom
                                                         Cristel Pelsser
                                               Internet Initiative Japan
                                                             Keyur Patel
                                                       Clarence Filsfils
                                                           Cisco Systems
                                                        October 26, 2009 25, 2010

                     Graceful BGP session shutdown
                      draft-ietf-grow-bgp-gshut-01
                      draft-ietf-grow-bgp-gshut-02

Abstract

   This draft describes operational procedures aimed at reducing the
   amount of traffic lost during planned maintenances of routers,
   involving the shutdown of BGP peering sessions.

Status of this Memo

   This Internet-Draft is submitted to IETF in full conformance with the
   provisions of BCP 78 and BCP 79.  This document may contain material
   from IETF Documents or IETF Contributions published or made publicly
   available before November 10, 2008.  The person(s) controlling the
   copyright in some of this material may not have granted the IETF
   Trust the right to allow modifications of such material outside the
   IETF Standards Process.  Without obtaining an adequate license from
   the person(s) controlling the copyright in such materials, this
   document may not be modified outside the IETF Standards Process, and
   derivative works of it may not be created outside the IETF Standards
   Process, except to format it for publication as an RFC or to
   translate it into languages other than English.

   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 April 29, 2010. 28, 2011.

Copyright Notice

   Copyright (c) 2009 2010 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
   (http://trustee.ietf.org/license-info) in effect on the date of
   publication of this document (http://trustee.ietf.org/license-info). document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.

Abstract  Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the BSD License.

   This draft describes operational procedures aimed at reducing document may contain material from IETF Documents or IETF
   Contributions published or made publicly available before November
   10, 2008.  The person(s) controlling the
   amount of traffic lost during planned maintenances copyright in some of routers,
   involving this
   material may not have granted the shutdown IETF Trust the right to allow
   modifications of BGP peering sessions.

Table such material outside the IETF Standards Process.
   Without obtaining an adequate license from the person(s) controlling
   the copyright in such materials, this document may not be modified
   outside the IETF Standards Process, and derivative works of Contents

   1. it may
   not be created outside the IETF Standards Process, except to format
   it for publication as an RFC or to translate it into languages other
   than English.

Table of Contents

   1.  Introduction . . . . . . . . . . . . . . . . . . . . . . . . .  4  3
   2.  Terminology  . . . . . . . . . . . . . . . . . . . . . . . . .  4  3
   3.  Packet loss upon manual eBGP session shutdown  . . . . . . . .  5  4
   4.  Practices to avoid packet losses . . . . . . . . . . . . . . .  5  4
     4.1.  Improving availability of alternate paths  . . . . . . . .  6  5
     4.2.  Graceful shutdown procedures for eBGP sessions . . . . . .  6  5
       4.2.1.  Outbound traffic . . . . . . . . . . . . . . . . . . .  6  5
       4.2.2.  Inbound traffic  . . . . . . . . . . . . . . . . . . .  7  6
       4.2.3.  Summary of operations  . . . . . . . . . . . . . . . .  8
       4.2.4.  BGP implementation support for G-Shut  . . . . . . . .  9
     4.3.  Graceful shutdown procedures for iBGP sessions . . . . . .  9
   5.  Forwarding modes and forwarding loops  . . . . . . . . . . . . 10
   6.  Dealing with Internet policies . . . . . . . . . . . . . . . . 10
   7.  Link Up cases  . . . . . . . . . . . . . . . . . . . . . . . . 11
     7.1.  Unreachability local to the ASBR . . . . . . . . . . . . . 11
     7.2.  iBGP convergence . . . . . . . . . . . . . . . . . . . . . 11
   8.  IANA considerations  . . . . . . . . . . . . . . . . . . . . . 12
   9.  Security Considerations  . . . . . . . . . . . . . . . . . . . 12
   10. Acknowledgments  . . . . . . . . . . . . . . . . . . . . . . . 13
   11. References . . . . . . . . . . . . . . . . . . . . . . . . . . 13
   Appendix A.  Summary of operations . . . . . . . . . . . . . . . . 13
     A.1.  Pre-configuration  . . . . . . . . . . . . . . . . . . . . 13
     A.2.  Operations at maintenance time . . . . . . . . . . . . . . 14
   Appendix B.  Alternative techniques with limited applicability . . 14
     B.1.
     A.1.  In-filter reconfiguration  . . . . . . . . . . . . . . . . 14
     B.2.
     A.2.  Multi Exit Discriminator tweaking  . . . . . . . . . . . . 15
     B.3.
     A.3.  IGP distance Poisoning . . . . . . . . . . . . . . . . . . 16
   Appendix C.  Effect of the g-shut procedure on the convergence . . 16
     C.1.  Maintenance of an eBGP session . . . . . . . . . 15
   Authors' Addresses . . . . . 16
       C.1.1.  Propagation on the other eBGP sessions of the
               g-shut initiator . . . . . . . . . . . . . . . . . . . 16
       C.1.2.  Propagation on 15

1.  Introduction

   Routing changes in BGP can be caused by planned, manual, maintenance
   operations.  This document discusses operational procedures to be
   applied in order to reduce or eliminate losses of packets during the other iBGP sessions
   maintenance.  These losses come from the transient lack of
   reachability during the
               g-shut initiator . . . . . . . . . . . . . . . . . . . 16
       C.1.3.  Propagation of updates in an iBGP full-mesh  . . . . . 17
       C.1.4.  Propagation of updates from iBGP to iBGP in a RR
               hierarchy  . . . . . . . . . . . . . . . . . . . . . . 17
     C.2.  Maintenance of an iBGP session . . . . . . . . . . . . . . 18
   Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 18

1.  Introduction

   Routing changes in BGP can be caused by planned, manual, maintenance
   operations.  This document discusses operational procedures to be
   applied in order to reduce or eliminate losses of packets during the
   maintenance.  These losses come from the transient lack of
   reachability during the BGP convergence following the shutdown BGP convergence following the shutdown of an
   eBGP peering session between two Autonomous System Border Routers
   (ASBR).

   This document presents procedures for the cases where the forwarding
   plane is impacted by the maintenance, hence when the use of Graceful
   Restart does not apply.

   The procedures described in this document can be applied to reduce or
   avoid packet loss for outbound and inbound traffic flows initially
   forwarded along the peering link to be shut down.  These procedures
   allow routers to keep using old paths until alternate ones are
   learned, ensuring that routers always have a valid route available
   during the convergence process.

   The goal of the document is to meet the requirements described in
   [REQS] at best, without changing the BGP protocol or BGP
   implementations.

   Still, it explains why reserving a community value for the purpose of
   BGP session graceful shutdown would reduce the management overhead
   bound with the solution.  It would also allow vendors to provide an
   automatic graceful shutdown mechanism that does not require any
   router reconfiguration at maintenance time.

   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].

2.  Terminology

   g-shut initiator : a router on which the session shutdown is
   performed for the maintenance.

   g-shut neighbor : a router that peers with the g-shut initiator via
   (one of) the session(s) to be shut down.

   Note that for the link-up case, we will refer to these nodes as g-no-
   shut initiator, and g-no-shut neighbor.

   Initiator AS : the Autonomous System of the g-shut initiator.

   Neighbor AS : the Autonomous System of the g-shut neighbor.

   Affected path / Nominal / pre-convergence path : a BGP path via the
   peering link(s) undergoing the maintenance.  This path will no longer
   exist after the shutdown.

   Affected prefix : a prefix initially reached via an affected path.

   Affected router : a router having an affected prefix.

   Backup / alternate / post-convergence path : a path towards an
   affected prefix that will be selected as the best path by an affected
   router, when the link is shut down and the BGP convergence is
   completed.

   Transient alternate path : a path towards an affected prefix that may
   be transiently selected as best by an affected router during the
   convergence process but that is not a post-convergence path.

   Loss of Connectivity (LoC) : the state when a router has no path
   towards an affected prefix.

3.  Packet loss upon manual eBGP session shutdown

   Packets can be lost during a manual shutdown of an eBGP session for
   two reasons.

   First, routers involved in the convergence process can transiently
   lack of paths towards an affected prefix, and drop traffic destined
   to this prefix.  This is because alternate paths can be hidden by
   nodes of an AS.  This happens when the paths are not selected as best
   by the ASBR that receive them on an eBGP session, or by Route
   Reflectors that do not propagate them further in the iBGP topology
   because they do not select them as best.

   Second, within the AS, routers' the FIB of routers can be transiently
   inconsistent during the BGP convergence and packets towards affected
   prefixes can loop and be dropped.  Note that these loops only happen
   when ASBR-to-
   ASBR ASBR-to-ASBR encapsulation is not used within the AS.

   This document only addresses the first reason.

4.  Practices to avoid packet losses

   This section describes means for an ISP to reduce the transient loss
   of packets upon a manual shutdown of a BGP session.

4.1.  Improving availability of alternate paths

   All solutions that increase the availability of alternate BGP paths
   in
   at routers performing packet lookups in BGP tables [BestExternal]
   [AddPath] help in reducing the LoC bound with manual shutdown of eBGP
   sessions.

   One of such solutions increasing diversity in such a way that, at any
   single step of the convergence process following the eBGP session
   shutdown, a BGP router does not receive a message withdrawing the
   only path it currently knows for a given NLRI, allows for a
   simplified g-shut procedure.  This simplified procedure would only
   tackle potential LoC for

   Increasing diversity with [AddPath] might lead to the inbound traffic. respect of this
   property, depending on the path propagation decision process that
   add-path compliant routers would use.

   Using advertise-best-external [BestExternal] on ASBRs and RRs helps
   in avoiding lack of alternate paths in route reflectors upon a
   convergence.  Hence it reduces the LoC duration for the outbound
   traffic of the ISP upon an eBGP Session shutdown by reducing the iBGP
   path hunting.

   Still it does not ensure that BGP routers will always have at least
   one path towards affected prefixes during the convergence following
   the event.  This property may be verified in future revisions of
   [BestExternal], notably of its Section 3, hence the current proposal
   will be updated accordingly.

   Increasing diversity with [AddPath] might lead to the respect of this
   property, depending on the path propagation decision process that
   add-path compliant routers would use.

   Note that the LoC for the inbound traffic of the maintained router,
   induced by a lack of alternate path propagation within the iBGP
   topology of a neighboring AS is not under the control of the operator
   performing the maintenance, hence the maintenance.  The procedure described in Section 4.2.2
   should thus be applied upon the maintenance, even if not
   required for the outbound traffic. procedure
   described in Section 4.2.1 is not applied.

4.2.  Graceful shutdown procedures for eBGP sessions

   This section aims at describing a procedure to be applied to reduce
   the LoC with readily available BGP features, and without assuming a
   particular iBGP design in the Initiator and Neighbor ASes.

4.2.1.  Outbound traffic

   This section discusses a mean to render the affected paths less
   desirable by the BGP decision process of affected routers, still
   allowing these to be used during the convergence convergence, while alternate
   paths are propagated to the affected routers.

   A decrease of the local-pref value of the affected paths can be
   issued in order to render the affected paths less preferable, at the
   highest possible level of the BGP Decision Process.

   This operation can be performed by reconfiguring the out-filters
   associated with the iBGP sessions established by the g-shut
   initiator.

   The modification of the filters MUST supplant any other rule
   affecting the local-pref value of the old paths.

   Compared to using an in-filter of the eBGP session to be shut down,
   the modification of the out-filters will not let the g-shut initiator
   switch to another path, as the input to the BGP decision process of
   that router does not change.  As a consequence, the g-shut initiator
   will not send a withdraw message over its iBGP modify the state of its dataplane, and will not withdraw the
   affected paths over its iBGP sessions when it receives an alternate path over an iBGP session.
   paths.  It will however modify the local-pref of the affected paths
   so that upstream routers will switch to alternate ones.

   When the actual shutdown of the session is performed, the g-shut
   initiator will itself switch to the alternate paths.

   In cases some BGP speakers in the AS override the local-pref
   attribute of paths received over iBGP sessions, the procedure
   described above will not work.  In such cases, the recommended
   procedure is to tag the paths sent over the iBGP sessions of the
   g-shut initiator with an AS specific community.  This AS specific
   community should lead to the setting of a low the lowest local-pref value.
   To be effective, the configuration related to this community MUST
   supplant or be applied after the already configured local-pref
   overriding.

   An operator may decide to follow a simplified procedure and directly
   apply an in-filter reducing the local preference of the paths
   received over the eBGP session being brought down.  While this
   procedure will be effective in many cases, corner cases as described
   in Appendix A.1 may happen, which may lead to some LoC for some
   affected destinations.  The use of this simplified procedure does not
   lead to LoC when used in conjunction with [BestExternal].

4.2.2.  Inbound traffic

   The solution described for the outbound traffic can be applied at the
   neighbor AS.  This can be done either "manually" or by using a
   community value dedicated to this task.

4.2.2.1.  Phone call

   The operator performing the maintenance of the eBGP session can
   contact the operator at the other side of the peering link, and let
   him apply the procedure described above for its own outbound traffic.

4.2.2.2.  Community tagging

   A community value (referred to as GSHUT community in this document)
   can be agreed upon by neighboring ASes.  A path tagged with this
   community must be considered as soon ASes and used to be affected by a maintenance
   operation. trigger the g-shut
   behavior at the g-shut neighbor.

4.2.2.2.1.  Pre-Configuration

   A g-shut neighbor is pre-configured to set a low local-pref value for
   the paths received over eBGP sessions which are tagged with the GSHUT
   community.

   This rule must supplant any other rule affecting the local-pref value
   of the paths.

   This local-pref reconfiguration SHOULD be performed at the out-
   filters of the iBGP sessions of the g-shut neighbor.  That is, the
   g-shut neighbor does not take into account this low local-pref in its
   own BGP best path selection.  As described in Section 4.2.1 this
   approach avoids sending the withdraw messages that can lead to LoC. LoC in
   some cases.

4.2.2.2.2.  Operational action upon maintenance

   Upon the manual shutdown, the output filter associated with the
   maintained eBGP session will be modified on the g-shut initiator so
   as to tag all the paths advertised over the session with the GSHUT
   community.

4.2.2.2.3.  Transitivity of the community

   If the GSHUT community is an extended community, it SHOULD be chosen
   non-transitive.  In that case, the clarification described in
   [Clarification4360] is required.

   If a regular community is used, this community SHOULD be removed from
   the path when the path is propagated over eBGP sessions.

   Not propagating the community further in the Internet reduces the
   amount of BGP churn and avoids rerouting in distant ASes that would
   also recognize this community value.  In other words, from a routing
   stability perspective, it helps concealing the convergence at the
   maintenance location.  From a security policy perspective, it prevents
   malignant ASes from using the community over paths propagated through
   intermediate ASes that do not support the feature, in order to
   perform inbound traffic engineering at the first AS recognizing the
   community.

   ASes which support the g-shut procedure SHOULD remove the community
   value(s) that they use for g-shut from the paths received from
   neighboring ASes that do not support the procedure or to whom the
   service is not provided.

   There are cases where an interdomain exploration is to be performed
   to recover the reachability, e.g., in the case of a shutdown in
   confederations where the alternate paths will be found in another AS
   of the confederation.  In such scenarios, the community value SHOULD
   be allowed to transit through the confederation but SHOULD be removed
   from the paths advertised outside of the confederation.

   When the local-pref value of a path is conserved upon its propagation
   from one AS of the confederation to the other, there is no need to
   have the GSHUT community be propagated throughout that confederation.

4.2.2.2.4.  Easing the configuration for G-SHUT

   From a configuration burden viewpoint, it is much easier to use a
   single dedicated value for the GSHUT community.

   First, on the g-shut initiator, an operator would have a single
   configuration rule to be applied at the maintenance time, which would
   not depend on the identity of its peer.  This would make the
   maintenance operations less error prone.

   Second, on the g-shut neighbor, a simple filter related to g-shut can
   be applied to all iBGP sessions.  Additionnaly, this filter does not
   need to be updated each time neighboring ASes are added or removed.

   The FCFS community value 0xFFFF0000 has been reserved for this
   purpose [BGPWKC].

4.3.  Graceful shutdown procedures for iBGP sessions

   If the iBGP topology is viable after the maintenance

4.2.3.  Summary of operations

   This section summarizes the session,
   i.e, if all BGP speakers of configurations and actions to be
   performed to support the AS have an iBGP signaling path for
   all prefixes advertised on this g-shut iBGP session, then procedure for eBGP peering links.

4.2.3.1.  Pre-configuration

   On each ASBR supporting the
   shutdown of g-shut procedure, set-up an out-filter
   applied on all iBGP session does not lead to transient
   unreachability.

   However, in the case of a shutdown of a router, a reconfiguration sessions of the out-filters ASBR, that :
   o    sets the local-pref of the paths tagged with the g-shut initiator MAY be performed
        community to set a low
   local-pref value for the paths originated by
   o    removes the g-shut initiator
   (e.g, BGP aggregates redistributed community from other protocols, including
   static routes).

   This behavior is equivalent to the recommended behavior for paths.
   o    optionally, adds an AS specific g-shut community on these paths
   "redistributed" from eBGP sessions
        to iBGP sessions in the case of
   the shutdown of an ASBR.

5.  Forwarding modes and forwarding loops indicate that these are to be withdrawn soon.  If some
        ingress ASBRs reset the local preference attribute, this AS applying
        specific g-shut community will be used to override other local
        preference changes.

4.2.3.2.  Operations at maintenance time

   On the solution does not rely g-shut initiator :
   o  Apply an out-filter on encapsulation to
   forward packets from the Ingress Border Router maintained eBGP session to tag the Egress Border
   Router, then transient forwarding loops and consequent packet losses
   can occur during the convergence process, even if
      paths propagated over the procedure
   described above is applied.  Hence if zero LoC is required,
   encapsulation is required between ASBRs of session with the AS.

   Using g-shut community.
   o  Apply an in-filter on the out-filter reconfiguration avoids maintained eBGP session to tag the forwarding loops
   between paths
      received over the g-shut initiator and its directly connected upstream
   neighboring routers.  Indeed, when this reconfiguration is applied, session with the g-shut initiator keeps using its own external path and lets the
   upstream routers converge community.
   o  Wait for convergence to happen.
   o  Perform a BGP session shutdown.

4.2.4.  BGP implementation support for G-Shut

   A BGP router implementation MAY provide features aimed at automating
   the alternate ones.  During this phase,
   no forwarding loops can occur between application of the g-shut initiator and its
   upstream neighbors graceful shutdown procedures described above.

   Upon a session shutdown specified as to be graceful by the operator,
   a BGP implementation supporting a g-shut initiator keeps using feature would

   1.   Update all the affected paths via its propagated over the corresponding eBGP peering links.  When all
        session, tagging the upstream routers have
   switched GSHUT community to them.  Any subsequent
        update sent to alternate paths, the transition performed by session being gracefully shut down would be
        tagged with the g-shut
   initiator when GSHUT community.
   2.   Lower the local preference value of the paths received over the
        eBGP session is actually being shut down, will be loopfree.
   Transient forwarding loops between other routers will not be avoided
   with this procedure.

6.  Dealing upon their propagation over iBGP
        sessions.  Optionally, also tag these paths with Internet policies

   A side gain an AS specific
        g-shut community.  Note that alternatively, the local preference
        of the maintenance solution is that it paths received over the eBGP session can also reduce be lowered on
        the churn implied by a shutdown g-shut initiator itself, instead of an eBGP session.

   For this, it is recommended only when propagating
        over its iBGP sessions.  This simplified behavior can lead to apply the filters modifying
        some LoC, as described in Appendix A.1, if not used in
        conjunction with [BestExternal].
   3.   Optionally shut down the local-
   pref value of session after a configured time.
   4.   Prevent the GSHUT community from being inherited by a path that
        would aggregate some paths tagged with the GSHUT community.
        This behavior avoids the GSHUT procedure to values strictly lower but as close as
   possible be applied to the local-pref values of
        aggregate upon the post-convergence paths.

   For example, if an eBGP link is shut down between a provider and graceful shutdown of one of its customers, and another link with this customer remains active,
   then covered
        prefixes.

4.3.  Graceful shutdown procedures for iBGP sessions

   If the value of iBGP topology is viable after the local-pref maintenance of the old paths SHOULD be decreased
   to session,
   i.e, if all BGP speakers of the smallest possible value of AS have an iBGP signaling path for
   all prefixes advertised on this g-shut iBGP session, then the 'customer' local_pref range,
   minus 1.  Thus, routers will
   shutdown of an iBGP session does not transiently switch lead to transient
   unreachability.

   However, in the case of a shutdown of a router, a reconfiguration of
   the out-filters of the g-shut initiator MAY be performed to set a low
   local-pref value for the paths received originated by the g-shut initiator
   (e.g, BGP aggregates redistributed from shared-cost peers or providers, which could lead other protocols, including
   static routes).

   This behavior is equivalent to the
   propagation of withdraw messages over recommended behavior for paths
   "redistributed" from eBGP sessions with shared-cost
   peers and providers.

   Proceeding like this reduces both BGP churn and traffic shifting as
   routers will less likely switch to transient paths.

   In the above example, it also prevents transient unreachabilities iBGP sessions in the neighboring case of
   the shutdown of an ASBR.

5.  Forwarding modes and forwarding loops

   If the AS that are due applying the solution does not rely on encapsulation to
   forward packets from the sending of "abrupt" withdraw
   messages Ingress Border Router to shared-cost peers and providers.

7.  Link Up cases

   We identify two potential causes for the Egress Border
   Router, then transient forwarding loops and consequent packet losses upon an
   eBGP link up event.  The first one is local to
   can occur during the g-no-shut
   initiator, convergence process, even if the second one procedure
   described above is due to applied.  Hence if zero LoC is required,
   encapsulation is required between ASBRs of the BGP convergence following AS.

   Using the
   injection of new best paths within out-filter reconfiguration avoids the iBGP topology.

7.1.  Unreachability local to forwarding loops
   between the ASBR

   An ASBR that selects as best a path received over a newly brought up
   eBGP session may transiently drop traffic.  This can typically happen g-shut initiator and its directly connected upstream
   neighboring routers.  Indeed, when this reconfiguration is applied,
   the nexthop attribute differs from g-shut initiator keeps using its own external path and lets the IP address of
   upstream routers converge to the eBGP
   peer, alternate ones.  During this phase,
   no forwarding loops can occur between the g-shut initiator and its
   upstream neighbors as the receiving ASBR has not yet resolved g-shut initiator keeps using the MAC address
   associated with affected
   paths via its eBGP peering links.  When all the IP address of that "third party" nexthop.

   A BGP speaker implementation could avoid such losses upstream routers have
   switched to alternate paths, the transition performed by ensuring that
   "third party" nexthops are resolved before installing paths using
   these in the RIB.

   If g-shut
   initiator when the link up event corresponds to an eBGP session that is being
   manually brought up, over an already up multi-access link, then actually shut down, will be loopfree.
   Transient forwarding loops between other routers will not be avoided
   with this procedure.

6.  Dealing with Internet policies

   A side gain of the
   operator can ping third party nexthops maintenance solution is that are expected it can also reduce
   the churn implied by a shutdown of an eBGP session.

   For this, it is recommended to be used
   before actually bringing apply the session up, or ping directed broadcast filters modifying the subnet IP address local-
   pref value of the link.  By proceeding like this, the MAC
   addresses associated with these third party nexthops will be resolved
   by the g-no-shut initiator.

7.2.  iBGP convergence

   Similar corner cases paths to values strictly lower but as described in Appendix C.1.4 for close as
   possible to the link down
   case, can occur during local-pref values of the post-convergence paths.

   For example, if an eBGP link up event.

   A typical example for such transient unreachability for a given
   prefix is the following :

        1.  A Route Reflector, RR1, is initially advertising the current
        best path to the members of its iBGP RR full-mesh.  It
        propagated that path within its RR full-mesh.  Another route
        reflector of the full-mesh, RR2, knows only that path towards
        the prefix.
        2.  A third Route Reflector of the RR full-mesh, RR3 receives shut down between a
        new best path orginated by the "g-no-shut" initiator, being provider and one
   of its RR clients.  RR3 selects it as best, and propagates an
        UPDATE within its RR full-mesh, i.e., to RR1 and RR2.
        3.  RR1 receives that path, reruns its decision process, customers, and
        picks another link with this new path as best.  As a result, RR1 withdraws its
        previously announced best-path on customer remains active,
   then the iBGP sessions value of its RR
        full-mesh.
        4.  If, for any reason, RR3 processes the withdraw generated in
        step 3, before processing local-pref of the update generated in step 2, RR3 old paths SHOULD be decreased
   to the smallest possible value of the 'customer' local_pref range,
   minus 1.  Thus, routers will not transiently suffers switch to paths received
   from unreachability for shared-cost peers or providers, which could lead to the affected prefix.

   The use
   propagation of [BestExternal] among withdraw messages over eBGP sessions with shared-cost
   peers and providers.

   Proceeding like this reduces both BGP churn and traffic shifting as
   routers will less likely switch to transient paths.

   In the RR of above example, it also prevents transient unreachabilities in
   the iBGP full-mesh can
   solve these corner cases by ensuring neighboring AS that within an AS, are due to the
   advertisement sending of a new route "abrupt" withdraw
   messages to shared-cost peers and providers.

7.  Link Up cases

   We identify two potential causes for transient packet losses upon an
   eBGP link up event.  The first one is not translated into local to the withdraw g-no-shut
   initiator, the second one is due to the BGP convergence following the
   injection of a
   former route.

   Indeed, "best-external" ensures that an ASBR does not withdraw a
   previously advertised (eBGP) path when it receives an additional,
   preferred path over an new best paths within the iBGP session.  Also, "best-intra-cluster"
   ensures topology.

7.1.  Unreachability local to the ASBR

   An ASBR that selects as best a RR does not withdraw a previously advertised (iBGP) path to its non clients (e.g. other RRs in received over a mesh of RR) newly brought up
   eBGP session may transiently drop traffic.  This can typically happen
   when it
   receives a new, preferred path over an iBGP session.

8.  IANA considerations

   Applying the g-shut procedure is rendered much easier with a reserved
   g-shut community value.  The community value 0xFFFF0000 has been
   reserved nexthop attribute differs from the FCFS community pool for this purpose.

9.  Security Considerations

   By providing IP address of the g-shut service to a neighboring AS, an ISP provides
   means to this neighbor to lower eBGP
   peer, and the local-pref value assigned to receiving ASBR has not yet resolved the
   paths received from this neighbor.

   The neighbor could abuse MAC address
   associated with the technique and do inbound traffic
   engineering IP address of that "third party" nexthop.

   A BGP speaker implementation could avoid such losses by declaring some prefixes as undergoing a maintenance so
   as to switch traffic to another peering link. ensuring that
   "third party" nexthops are resolved before installing paths using
   these in the RIB.

   If this behavior the link up event corresponds to an eBGP session that is not tolerated by being
   manually brought up, over an already up multi-access link, then the ISP, it SHOULD monitor
   operator can ping third party nexthops that are expected to be used
   before actually bringing the
   use session up, or ping directed broadcast
   the subnet IP address of the g-shut community link.  By proceeding like this, the MAC
   addresses associated with these third party nexthops will be resolved
   by this neighbor.

   ASes which support the g-shut procedure SHOULD remove g-no-shut initiator.

7.2.  iBGP convergence

   Similar corner cases as described in Appendix A.1 for the community
   value(s) that they use link down
   case, can occur during an eBGP link up event.

   A typical example for g-shut from such transient unreachability for a given
   prefix is the paths received from
   neighboring ASes that do not support following :

        1.  A Route Reflector, RR1, is initially advertising the procedure or current
        best path to whom the
   service is not provided.  Doing so prevents malignant ASes from using the community through intermediate ASes members of its iBGP RR full-mesh.  It
        propagated that do not support the
   feature, in order to perform inbound traffic engineering.

10.  Acknowledgments

   The authors wish to thank Olivier Bonaventure and Pradosh Mohapatra
   for their useful comments on this work.

11.  References

   [AddPath]  D. Walton, A. Retana, and E. Chen, "Advertisement of
              Multiple Paths in BGP", draft-walton-bgp-add-paths-06.txt
              (work in progress).

   [BestExternal]
              Marques, P., Fernando, R., Chen, E., and P. Mohapatra,
              "Advertisement of the best-external route to IBGP",
               draft-ietf-idr-best-external-00.txt, May 2009.

   [REQS]     Decraene, B., Francois, P., Pelsser, C., Ahmad, Z., and T.
              Takeda, "Requirements for the graceful shutdown of BGP
              sessions",
               draft-ietf-grow-bgp-graceful-shutdown-requirements-
              01.txt, October 2009.

   [RFC4360]  Sangli, S., Tappan, D., and Y. Rekhter, "BGP Extended
              Communities Attribute", RFC 4360, February 2006.

   [Clarification4360]
              Decraene, B., Vanbever, L., and P. Francois, "RFC 4360
              Clarification Request",
               draft-decraene-idr-rfc4360-clarification-00,
              October 2009.

   [BGPWKC]   "http://www.iana.org/assignments/
              bgp-well-known-communities".

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119, March 1997.

Appendix A.  Summary of operations

   This section summarizes the configurations and actions to be
   performed to support the g-shut procedure for eBGP peering links.

A.1.  Pre-configuration

   On each ASBR supporting the g-shut procedure, set-up an out-filter
   applied on all iBGP sessions of the ASBR, that :

   . sets the local-pref of the paths tagged with the g-shut community
   to a low value

   . removes the g-shut community from the path.

   Optionally, add an AS specific g-shut community on the path to
   indicate that this path is to be shutdown.  If some ingress ASBRs
   reset the local preference attribute, this AS specific g-shut
   community will be used to override other local preference changes.

A.2.  Operations at maintenance time

   On the g-shut initiator :

   .  Apply an in-filter on the maintained eBGP session to tag the paths
   received over the session with the g-shut community.

   .  Apply an out-filter on the maintained eBGP session to tag the
   paths propagated over the session with the g-shut community.

   .  Wait for convergence to happen.

   .  Perform a BGP session shutdown.

Appendix B.  Alternative techniques with limited applicability

   A few alternative techniques have been considered to provide g-shut
   capabilities but have been rejected due to their limited
   applicability.  This section describe them for possible reference.

B.1.  In-filter reconfiguration

   An In-filter reconfiguration on the eBGP session undergoing the
   maintenance could be performed instead of out-filter reconfigurations
   on the iBGP sessions of the g-shut initiator.

   Upon the application of the maintenance procedure, if the g-shut
   initiator has an alternate path in its Adj-Rib-In, it will switch to
   it directly.

   If this new path was advertised by an eBGP neighbor of the g-shut
   initiator, the g-shut initiator will send a BGP Path Update message
   advertising the new path over its iBGP and eBGP sessions.

   If this new path was received over an iBGP session, the g-shut
   initiator will select that path and withdraw the previously
   advertised path over its non-client iBGP sessions.  There can be iBGP
   topologies where the iBGP peers of the g-shut initiator do not know
   an alternate path, and hence may drop traffic.

   Also, applying an In-filter reconfiguration on the eBGP session
   undergoing the maintenance may lead to transient LoC, in full-mesh
   iBGP topologies if

        a.  An ASBR of the initiator AS, ASBR1 did not initially select within its own external path as best, and

        b.  An ASBR RR full-mesh.  Another route
        reflector of the initiator AS, ASBR2 advertises a new full-mesh, RR2, knows only that path
        along its iBGP sessions upon towards
        the reception prefix.
        2.  A third Route Reflector of ASBR1's update
        following the in-filter reconfiguration on RR full-mesh, RR3 receives a
        new best path orginated by the g-shut "g-no-shut" initiator, being one
        of its RR clients.  RR3 selects it as best, and

        c.  ASBR1 receives the update message, runs propagates an
        UPDATE within its Decision Process RR full-mesh, i.e., to RR1 and hence withdraws RR2.
        3.  RR1 receives that path, reruns its external path after having selected
        ASBR2's decision process, and
        picks this new path as best, and

        d.  An impacted router of best.  As a result, RR1 withdraws its
        previously announced best-path on the AS iBGP sessions of its RR
        full-mesh.
        4.  If, for any reason, RR3 processes the withdraw of ASBR1 generated in
        step 3, before processing the update generated in step 2, RR3
        transiently suffers from ASBR2.

   Applying a reconfiguration unreachability for the affected prefix.

   The use of [BestExternal] among the out-filters prevents such transient
   unreachabilities.

   Indeed, when RR of the g-shut initiator propagates iBGP full-mesh can
   solve these corner cases by ensuring that within an update of AS, the old
   path first,
   advertisement of a new route is not translated into the withdraw from ASBR2 of a
   former route.

   Indeed, "best-external" ensures that an ASBR does not trigger unreachability
   in other nodes, as the old withdraw a
   previously advertised (eBGP) path is still available.  Indeed, even
   though when it receives an additional,
   preferred path over an iBGP session.  Also, "best-intra-cluster"
   ensures that a RR does not withdraw a previously advertised (iBGP)
   path to its non clients (e.g. other RRs in a mesh of RR) when it
   receives alternate paths, the g-shut initiator keeps using
   its old a new, preferred path as best as the in-filter of the maintained eBGP session
   has not been modified yet. over an iBGP session.

8.  IANA considerations

   Applying the out-filter reconfiguration also prevents packet loops
   between the g-shut initiator and its direct neighbors when
   encapsulation procedure is not used between the ASBRs of the AS.

B.2.  Multi Exit Discriminator tweaking rendered much easier with a reserved
   g-shut community value.  The MED attribute of community value 0xFFFF0000 has been
   reserved from the paths FCFS community pool for this purpose.

9.  Security Considerations

   By providing the g-shut service to be avoided can be increased so as a neighboring AS, an ISP provides
   means to force the routers in this neighbor to lower the neighboring AS local-pref value assigned to select other paths.

   The solution only works if the alternate
   paths are received from this neighbor.

   The neighbor could abuse the technique and do inbound traffic
   engineering by declaring some prefixes as good undergoing a maintenance so
   as the
   initial ones with respect to the Local-Pref value and the AS Path
   Length value.  In the other cases, increasing the MED value will switch traffic to another peering link.

   If this behavior is not
   have an impact on tolerated by the decision process of ISP, it SHOULD monitor the routers in
   use of the
   neighboring AS.

B.3.  IGP distance Poisoning

   The distance to g-shut community by this neighbor.

   ASes which support the BGP nexthop corresponding to g-shut procedure SHOULD remove the maintained
   session can be increased in community
   value(s) that they use for g-shut from the IGP so paths received from
   neighboring ASes that do not support the old paths will be
   less preferred during the application of procedure or to whom the IGP distance tie-break
   rule.  However, this solution only works for
   service is not provided.  Doing so prevents malignant ASes from using
   the paths whose
   alternates are as good as community through intermediate ASes that do not support the old paths with respect
   feature, in order to their Local-
   Pref value, their AS Path length, perform inbound traffic engineering.

10.  Acknowledgments

   The authors wish to thank Olivier Bonaventure and Pradosh Mohapatra
   for their MED value.

   Also, useful comments on this poisoning cannot be applied when nexthop self is used as
   there is no nexthop specific to the maintained session to poison work.

11.  References

   [AddPath]  D. Walton, A. Retana, and E. Chen, "Advertisement of
              Multiple Paths in
   the IGP.

Appendix C.  Effect BGP", draft-walton-bgp-add-paths-06.txt
              (work in progress).

   [BestExternal]
              Marques, P., Fernando, R., Chen, E., and P. Mohapatra,
              "Advertisement of the g-shut procedure on the convergence

   This section describes the effect of applying best-external route to IBGP",
               draft-ietf-idr-best-external-00.txt, May 2009.

   [REQS]     Decraene, B., Francois, P., Pelsser, C., Ahmad, Z.,
              Armengol, A., and T. Takeda, "Requirements for the solution.

C.1.  Maintenance
              graceful shutdown of an eBGP session BGP sessions",
               draft-ietf-grow-bgp-graceful-shutdown-requirements-
              06.txt, October 2010.

   [RFC4360]  Sangli, S., Tappan, D., and Y. Rekhter, "BGP Extended
              Communities Attribute", RFC 4360, February 2006.

   [Clarification4360]
              Decraene, B., Vanbever, L., and P. Francois, "RFC 4360
              Clarification Request",
               draft-decraene-idr-rfc4360-clarification-00,
              October 2009.

   [BGPWKC]   "http://www.iana.org/assignments/
              bgp-well-known-communities".

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119, March 1997.

Appendix A.  Alternative techniques with limited applicability

   A few alternative techniques have been considered to provide g-shut
   capabilities but have been rejected due to their limited
   applicability.  This section describes the effect of applying the solution describe them for the
   shutdown of an eBGP session.

C.1.1.  Propagation on the other eBGP sessions of the g-shut initiator

   Nothing is propagated on the other eBGP sessions when the out-filters
   reconfiguration step is applied.  The possible reference.

A.1.  In-filter reconfiguration is indeed only
   defined for its iBGP sessions.

   The

   An In-filter reconfiguration of the iBGP out-filters will trigger the
   reception of alternate paths at the g-shut initiator.  As on the eBGP
   in-filters have not been modified at that step, the old paths are
   still preferred by the g-shut initiator.

C.1.2.  Propagation on session undergoing the other iBGP sessions
   maintenance could be performed instead of the g-shut initiator

   During the out-filter reconfiguration, path updates are propagated
   with a reduced local-pref value for the affected paths.  As a
   consequence, Route Reflectors and distant ASBRs select and propagate
   alternate paths through the iBGP topology as they no longer select
   the old paths as best.

   When the shut-down is performed, for each affected prefix, the g-shut
   initiator propagates reconfigurations
   on its iBGP sessions:

   .  The alternate path, if the best path was received over an eBGP
   sessions.

   .  A withdraw, if the best path was received over an iBGP sessions.

C.1.3.  Propagation of updates in an iBGP full-mesh

   No transient LoC can occur if a reconfiguration sessions of the iBGP out-
   filters on the g-shut initiator is performed.

C.1.4.  Propagation of updates from iBGP to iBGP in a RR hierarchy initiator.

   Upon the reception application of the update of a primary path with a lower
   local-pref value from a client, a Route Reflector RR1 will either
   propagate maintenance procedure, if the update, or select g-shut
   initiator has an alternate path, depending on the
   fact that the updated primary path is still in its Adj-Rib-In, it will switch to
   it directly.

   If this new path was advertised by an eBGP neighbor of the best one w.r.t. g-shut
   initiator, the
   state of g-shut initiator will send a BGP Path Update message
   advertising the Adj-Rib-In of RR1. new path over its iBGP and eBGP sessions.

   If the updated primary this new path is still the best, then was received over an iBGP session, the RR g-shut
   initiator will
   propagate an update for this select that path to and withdraw the iBGP neighbors to which it previously
   advertised the path.  Hence it cannot cause transient lack
   of path in the Adj-Rib-In of over its non-client iBGP neighbors.

   If sessions.  There can be iBGP
   topologies where the iBGP peers of the g-shut initiator do not know
   an alternate path is picked, path, and this path was also originated by
   a client of RR1, hence may drop traffic.

   Also, applying an update will also be propagated to In-filter reconfiguration on the same
   neighbors as eBGP session
   undergoing the one maintenance may lead to which the primary path was initially
   propagated.  Hence it cannot cause transient lack of path LoC, in the Adj-
   Rib-In full-mesh
   iBGP topologies if

        a.  An ASBR of the initiator AS, ASBR1 did not initially select
        its iBGP neighbors.

   If an alternate own external path is picked, as best, and this path was received from a
   member

        b.  An ASBR of the initiator AS, ASBR2 advertises a new path
        along its Route-Reflector iBGP full-mesh, then a withdraw message
   is sent.  As sessions upon the alternate reception of ASBR1's update
        following the in-filter reconfiguration on the g-shut initiator,
        and

        c.  ASBR1 receives the update message, runs its Decision Process
        and hence withdraws its external path has been sent over each session after having selected
        ASBR2's path as best, and

        d.  An impacted router of the iBGP full-mesh, AS processes the propagation of a withdraw for the primary
   path of RR1 is done to routers that are expected to know ASBR1
        before processing the
   alternate path picked by RR1.

   The following example describes a situation where some corner case
   timings could lead to transient unreachability update from some members ASBR2.

   Applying a reconfiguration of the iBGP full-mesh.

        1.  A Route Reflector RR1 only knew about the primary path upon out-filters prevents such transient
   unreachabilities.

   Indeed, when the shutdown.

        2.  A member of its RR full-mesh, RR2, g-shut initiator propagates an update of the old
   path with a lower local-pref.

        3.  Another member of its RR full-mesh, RR3 processes first, the
        update, selects an alternate path, and propagates an update withdraw from ASBR2 does not trigger unreachability
   in other nodes, as the mesh.

        4.  RR2 old path is still available.  Indeed, even
   though it receives the alternate path, selects it as best, and
        hence withdraws paths, the updated g-shut initiator keeps using
   its old path on as best as the iBGP sessions in-filter of the
        mesh.

        5.  If for any reason, RR1 receives maintained eBGP session
   has not been modified yet.

   Applying the out-filter reconfiguration also prevents packet loops
   between the g-shut initiator and processes its direct neighbors when
   encapsulation is not used between the withdraw
        generated ASBRs of the AS.

   Note that applying this simplified procedure in step 4 before processing conjunction with
   [BestExternal] does not lead to LoC.

A.2.  Multi Exit Discriminator tweaking

   The MED attribute of the update generated paths to be avoided can be increased so as
   to force the routers in
        step 3, RR1 transiently suffers from unreachability for the
        affected prefix. neighboring AS to select other paths.

   The solution only works if the alternate paths are as good as the
   initial ones with respect to the Local-Pref value and the AS Path
   Length value.  In such corner cases, the solution improves other cases, increasing the iBGP convergence
   behavior/LoC but does MED value will not ensure 0 packet loss, as we cannot define a
   simple solution relying only
   have an impact on a reconfiguration of the filters decision process of the g-shut initiator.  Improving the availability of alternate paths routers in Route Reflectors, using [BestExternal], or [AddPath], seems the
   neighboring AS.

A.3.  IGP distance Poisoning

   The distance to be the most pragmatic solution BGP nexthop corresponding to these corner cases.

   The use of [BestExternal] in the iBGP full-mesh between RRs maintained
   session can solve
   these corner cases by ensuring that within an AS, the advertisement
   of a new path is not translated into be increased in the withdraw of a former path.

   Indeed, "best-external" ensures that an ASBR does not withdraw a
   previously advertised (eBGP) path when it receives an additional,
   preferred path over an iBGP session.  Also, "best-intra-cluster"
   ensures IGP so that a RR does not withdraw a previously advertised (iBGP)
   path to its non clients (e.g. other RRs in a mesh of RR) when it
   receives a new, preferred path over an iBGP session.

C.2.  Maintenance of an iBGP session

   If the shutdown does not temper with old paths will be
   less preferred during the viability application of the iBGP
   topology, IGP distance tie-break
   rule.  However, this solution only works for the described procedure paths whose
   alternates are as good as the old paths with respect to their Local-
   Pref value, their AS Path length, and their MED value.

   Also, this poisoning cannot be applied when nexthop self is used as
   there is sufficient no nexthop specific to avoid LoC. the maintained session to poison in
   the IGP.

Authors' Addresses

   Pierre Francois
   Universite catholique de Louvain
   Place Ste Barbe, 2
   Louvain-la-Neuve  1348
   BE

   Email: pierre.francois@uclouvain.be
   URI:   http://inl.info.ucl.ac.be/pfr

   Bruno Decraene
   France Telecom
   38-40 rue du General Leclerc
   92794 Issi Moulineaux cedex 9
   FR

   Email: bruno.decraene@orange-ftgroup.com

   Cristel Pelsser
   Internet Initiative Japan
   Jinbocho Mitsui Bldg.
   1-105 Kanda Jinbo-cho
   Tokyo  101-0051
   JP

   Email: pelsser.cristel@iij.ad.jp

   Keyur Patel
   Cisco Systems
   170 West Tasman Dr
   San Jose, CA  95134
   US

   Email: keyupate@cisco.com

   Clarence Filsfils
   Cisco Systems
   De kleetlaan 6a
   Diegem  1831
   BE

   Email: cfilsfil@cisco.com