Internet Engineering Task Force                            D. Ceccarelli
Internet-Draft                                               D. Caviglia
Intended status: Informational                               F. Fondelli
Expires: February 19, 2010                                      Ericsson
                                                                M. Corsi
                                                                  Altran
                                                         A. D'Alessandro
                                                          Telecom Italia
                                                         August 18, 2009


            P2MP traffic protection in MPLS-TP ring topology
                 draft-ceccarelli-mpls-tp-p2mp-ring-01

Status of this Memo

   This Internet-Draft is submitted to IETF in full conformance with the
   provisions of BCP 78 and 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 February 19, 2010.

Copyright Notice

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




Ceccarelli, et al.      Expires February 19, 2010               [Page 1]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


Abstract

   Purpose of this ID is to describe requirements and possible solutions
   for point to multipoint (P2MP) traffic distribution over
   interconnected MPLS-TP rings.  The rationale for an ID on such a
   specific application is illustrated in the rest of the document.


Table of Contents

   1.  Introduction . . . . . . . . . . . . . . . . . . . . . . . . .  3
     1.1.  Background . . . . . . . . . . . . . . . . . . . . . . . .  3
     1.2.  Scope of this document . . . . . . . . . . . . . . . . . .  3
   2.  Conventions used in this document  . . . . . . . . . . . . . .  4
   3.  Problem Statement  . . . . . . . . . . . . . . . . . . . . . .  4
   4.  Requirements . . . . . . . . . . . . . . . . . . . . . . . . .  5
     4.1.  Topology . . . . . . . . . . . . . . . . . . . . . . . . .  5
     4.2.  Bandwidth Optimization . . . . . . . . . . . . . . . . . .  7
     4.3.  Backup LSP Optimization  . . . . . . . . . . . . . . . . .  7
     4.4.  Protection switching time  . . . . . . . . . . . . . . . .  7
     4.5.  Revertiveness  . . . . . . . . . . . . . . . . . . . . . .  8
     4.6.  Frequent protection switching prevention . . . . . . . . .  8
     4.7.  Manual commands  . . . . . . . . . . . . . . . . . . . . .  8
     4.8.  Protection Mechanisms  . . . . . . . . . . . . . . . . . .  8
   5.  Possible solutions for P2MP traffic distribution . . . . . . .  8
   6.  Proposed recovery methods  . . . . . . . . . . . . . . . . . .  8
     6.1.  Fast Rerouting (FRR) . . . . . . . . . . . . . . . . . . .  9
       6.1.1.  Fast Rerouting-Transport Profile (FRR-TP)  . . . . . .  9
     6.2.  Optimized Multicast Fast Rerouting (ROM-FRR) . . . . . . . 14
     6.3.  Bandwidth Utilization Comparison . . . . . . . . . . . . . 17
     6.4.  Multiple Failures Comparison . . . . . . . . . . . . . . . 18
   7.  Conclusions  . . . . . . . . . . . . . . . . . . . . . . . . . 18
   8.  Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 19
   9.  IANA Considerations  . . . . . . . . . . . . . . . . . . . . . 19
   10. Security Considerations  . . . . . . . . . . . . . . . . . . . 19
   11. References . . . . . . . . . . . . . . . . . . . . . . . . . . 19
     11.1. Normative References . . . . . . . . . . . . . . . . . . . 19
     11.2. Informational References . . . . . . . . . . . . . . . . . 20
   Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 20












Ceccarelli, et al.      Expires February 19, 2010               [Page 2]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


1.  Introduction

1.1.  Background

   MPLS-TP is a joint ITU-T/IETF effort to include an MPLS Transport
   Profile within the IETF MPLS architecture to support the capabilities
   and functionalities of a packet transport network as defined by
   ITU-T.  In the MPLS-TP requirements draft [DRAFT-JENKIS] it is
   highlighted that an MPLS-TP architecture must allow a smooth
   migration from legacy networks (e.g.  SONET/SDH) to packet transport
   networks and support, in a reliable way, the accelerating growth of
   packet-based services (such as VoIP, L2/L3 VPN, IPTV, RAN
   backhauling, etc.).  That migration must be accomplished preserving
   carriers investments in both Capital Expenditure (CAPEX) and
   Operational Expense (OPEX) as much as possible.  Most of today
   deployed SDH/SONET networks (some hundreds of thousands) all over the
   world are based on ring topology, this assumption being especially
   true for Metro-Core and Metro-Access networks.

   Metro Area ring based networks are usually composed by a main Metro-
   Core ring and several minor Metro-Access rings.  The interconnection
   between such rings is mainly based on a single node or on a couple of
   nodes (one or more hop away from the first one).

   Therefore, it is desirable that MPLS-TP solution was based on
   interconnected ring architectures that were previously used by SONET/
   SDH in order to avoid needs of digging new fibers (very expensive
   especially in Europe), or lease dark fibers, in metro areas and
   maintain low cost.  If we combine the previous topology
   considerations with the fact that the most interesting applications
   leading the network transformation are P2MP applications (e.g.
   IPTV), we reach the point that MPLS-TP must support an efficient
   solution for P2MP applications over interconnected rings.  It is also
   important to note that those high value applications need appropriate
   resiliency, at least single node/link failure recovery.

   The solutions proposed in this document are data plane driven and are
   thought to be able to work in absence of control plane.  Nevertheless
   a ring network allows the set up of bandwidth optimizing control
   plane driven solutions.  Such solutions are out of the scope of this
   document and will be discussed in further IDs.

1.2.  Scope of this document

   This document provides both functional requirements and a network
   solution for MPLS-TP ring based topology networks that support
   reliable point-to-multipoint services.




Ceccarelli, et al.      Expires February 19, 2010               [Page 3]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


2.  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].


3.  Problem Statement

   This document addresses the bandwidth usage optimization of reliable
   p2mp traffic distribution (e.g.  IPTV) over MPLS-TP networks based on
   interconnected ring topology (physical or logical).  Resiliency must
   be based on recovery mechanisms able to restore traffic in about
   50ms, without usage of any control plane accordingly with the MPLS-TP
   general requirements [DRAFT-JENKINS] the recovery mechanism should
   not rely on any control plane.  Control plane has to be intended in
   the sense of signaling and/or routing protocols, namely the GMPLS
   Control Plane suite.

   The following picture illustrates a typical interconnected ring
   topology that can be found in metro networks; of course real
   topologies involve a huger amount of rings and nodes.

   The source node can be considered as a video server that distributes
   a p2mp video stream towards different DSLAMs.  The usage of redundant
   video servers located in different nodes is allowed.  Reliable
   bandwidth optimization is for further study.

   DSLAM1 is directly connected to the MetroCore ring while DSLAM2 and
   DSLAM3 are connected to the Metro Access ring.





















Ceccarelli, et al.      Expires February 19, 2010               [Page 4]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


              Source
                 |
               +---+            +---+
               |   |------------|   |-->DSLAM1
               +---+            +---+
                 |                |
                 |   Metro Core   |
                 |      Ring      |
                 |                |
               +---+            +---+
               |   |------------|   |
               +---+            +---+
                /\
              /    \
            /        \      Metro Access Ring
         +---+      +---+
         |   |------|   |
         +---+      +---+
           |          |
           V          V
           DSLAM2     DSLAM3

                       Interconnected ring topology

                                 Figure 1


4.  Requirements

   This paragraph lists and explains the main requirements to be
   satisfied by a network in order to provide a reliable MPLS-TP based
   infrastructure for P2MP traffic transport.  P2MP traffic transport
   over an MPLS-TP network requirements are based on [DRAFT-JENKINS],
   [DRAFT-BLB] and [DRAFT-SPRECHER].

4.1.  Topology

   The network topology is made of one main ring (e.g.  Metro Core Ring)
   and more minor rings (e.g.  Metro Access Ring), if any.

   The interconnection between two rings can consist of one single node
   or a couple of nodes (the two nodes can be one or more hops away).
   The two configurations originate two different kinds of topology.  We
   will call those topologies "Single node ring interconnection" and
   "Dual node ring interconnection" respectively.






Ceccarelli, et al.      Expires February 19, 2010               [Page 5]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


                Source
                  |
                +---+
                |   |
                +---+
              /       \
             /         \
            /           \
         +---+         +---+
         |   |         |   |
         +---+         +---+
           |             |
           |             | Metro Core Ring
           |             |
         +---+         +---+
         |   |         |   |-->DSLAM1
         +---+         +---+
            \           /
             \         /
              \       /
                +---+
                |   |
                +---+
              /       \
             /         \
            /           \
         +---+         +---+
         |   |         |   |-->DSLAM2
         +---+         +---+
            \           /
             \         /   Metro Access Ring
              \       /
                +---+
                |   |
                +---+
                  |
                  V
                DSLAM3

                     Single node ring interconnection

                                 Figure 2









Ceccarelli, et al.      Expires February 19, 2010               [Page 6]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


              Source
                 |
               +---+            +---+
               |   |------------|   |-->DSLAM1
               +---+            +---+
                 |                |
                 |                |
                 |   METRO CORE   |
               +---+    RING    +---+
               |   |            |   |
               +---+            +---+
                 |                |
                 |                |
                 |                |
               +---+            +---+
               |   |------------|   |
               +---+            +---+
                 |                |
                 |  METRO ACCESS  |
                 |      RING      |
                 |                |
               +---+            +---+
               |   |------------|   |
               +---+            +---+
                 |                |
                 V                V
               DSLAM2          DSLAM3

                      Dual node ring interconnection

                                 Figure 3

4.2.  Bandwidth Optimization

   Each node MUST minimize packet replication and packets SHOULD not be
   replicated on the same unidirectional physical path.

4.3.  Backup LSP Optimization

   Optimization criteria and algorithms applied to the working LSP
   SHOULD be respected/applicable also to the backup LSPs when the
   protection is activated.

4.4.  Protection switching time

   The protection of the P2MP traffic over the interconnected rings MUST
   provide a switching time within 50 ms.  This means that the P2MP
   traffic must be recovered, in case of either link or node failure,



Ceccarelli, et al.      Expires February 19, 2010               [Page 7]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


   within 50ms from the fault detection.

4.5.  Revertiveness

   It MUST be possible to configure the protection switching in
   revertive or non revertive mode [DRAFT-BLB].

4.6.  Frequent protection switching prevention

   It MUST be possible to prevent frequent protection switching (i.e.
   Hold off and Wait To Restore timers).  [DRAFT-JENKINS]

4.7.  Manual commands

   Manual commands MUST be supported.  [DRAFT-JENKINS]

4.8.  Protection Mechanisms

   A set of resilience mechanisms MUST be available.  These mechanisms
   MUST NOT rely on the control plane.

   Protection mechanisms and control plane based resilient mechanisms
   MUST coexist.  [DRAFT-JENKINS]


5.  Possible solutions for P2MP traffic distribution

   The solutions proposed for the distribution of P2MP traffic over
   interconnected ring networks (single node ring interconnection or
   dual node ring interconnection), are based on P2MP LSPs [RFC
   3031][RFC4875] and P2MP PWE3s [DRAFT-PWE3-P2MP].

   The use of a P2MP LSP instead of many P2P LSPs reduces traffic
   replication and bandwidth utilization in the network.  However, the
   solution for P2MP LSP described in [RFC4875] must be enhanced in
   order to fully meet MPLS-TP P2MP ring distribution requirements.
   P2MP LSPs MUST be provisioned via management plane
   [RFC3812][RFC3813][P2MP-TE-MIB] without control plane support and
   they MUST provide protection switching mechanisms relying on MPLS-TP
   OAM in order to grant network survivability.


6.  Proposed recovery methods

   In this section two different recovery schemes based on the Fast
   Rerouting (FRR) technique are analyzed and compared.  The first
   scheme is called FRR-TP (Fast Rerouting - Transport Profile) because
   it is mainly based on the rerouting of the failed portion of the



Ceccarelli, et al.      Expires February 19, 2010               [Page 8]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


   network through a pre-provisioned path and exploits OAM
   functionalities.  The second one is called ROM-FRR (Ring Optimized
   Multicast - Fast ReRouting) due to the fact that, despite being
   applicable to any kind of network, it is optimized for the
   distribution of multicast traffic over ring networks.  The two
   different recovery schemes will be compared and pros and cons of each
   one will be highlighted.

6.1.  Fast Rerouting (FRR)

   [RFC4875] describes procedures to perform Fast Rerouting operations
   for P2MP LSPs using extensions defined in [RFC4090].

   The FRR mechanism is based on the re-direction of traffic flows on
   backup LSPs as soon as a failure is detected.  This switch is
   extremely fast due to the fact that the backup LSPs are computed and
   provisioned before the failure detection and the traffic is re-
   directed as close to the failure point as possible.  It is possible
   to reach switching times of tens of milliseconds because no path
   computation and set-up are performed and the failure notification
   must not be propagated to the ingress LER.

   In [RFC4090] two different methods are defined: one-to-one and
   facility backup.  The first method foreseen the creation of a detour
   LSP for each protected LSP at each potential point of local repair,
   while the second one creates a bypass tunnel to protect a set of LSPs
   with similar backup constraints.

   Both these methods are based on the assumption that each LSR of the
   protected LSP is a Merge Point (MP), that is, it must be able to
   rejoin the backup LSP to the protected one downstream of the
   potential failure.

6.1.1.  Fast Rerouting-Transport Profile (FRR-TP)

   Both one-to-one and facility backup can be used as recovery
   mechanisms for P2MP LSPs in MPLS-TP interconnected ring topology
   networks but they need to be extended in order to fully meet MPLS-TP
   protection requirements.  This solution will be called FRR-TP.

   MPLS-TP OAM SHOULD be used on each MPLS section to perform Continuity
   Check (CC) operations.  When a defect is detected, a hold-off timer
   (with period configurable from 1 to 10s in steps of 100ms) SHOLUD be
   started in order to prevent frequent protection switches.  When the
   timer expires and a defect condition is still present, the protection
   switching SHOULD be initiated.

   Externally initiated commands MAY be provided for manual control of



Ceccarelli, et al.      Expires February 19, 2010               [Page 9]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


   protection switching on each PLR.  The following externally initiated
   commands SHOULD be supported: Clear, Lockout of Protection, Forced
   Switch, Manual Switch, Exercise.

   In the following picture an example of FRR-TP application to a ring
   topology is depicted.


                                         Source
                                           |
                       +---+             +---+
               DSLAM3<-| F |-------------| A |
                       +---+             +---+
                        /                   \
                       /                     \
                      /                       \
                   +---+                     +---+
                   | E |                     | B |
                   +---+                     +---+
                      \                       /
                       \                     /
                        \                   /
                       +---+             +---+
                       | D |-------------| C |-->DSLAM1
                       +---+             +---+
                         |
                         V
                       DSLAM2

                    FRR-TP application to ring topology

                                 Figure 4

   The source of a P2MP traffic flow is connected to node A and a P2MP
   LSP is created with A as ingress node and C, D and F as egress nodes
   in order to deliver traffic to the different DSLAMs.  In the
   following figure it is possible to see the list of the protected LSP
   and all the possible backup LSPs in case of link failure and node
   failure configuration.  "X's Backup" is the backup path activated by
   X as a consequence of a failure affecting node Y (downstream node
   with respect to X) or link X-Y, and square brackets indicate nodes
   delivering traffic to the DSLAMs.









Ceccarelli, et al.      Expires February 19, 2010              [Page 10]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


                   Protected LSP:  A->B->[C]->[D]->E->[F]

                   --- LINK PROTECTION ---

                   A's Backup:     A->F->E->D->C->B
                   B's Backup:     B->A->F->E->D->C
                   C's Backup:     C->B->A->F->E->D
                   D's Backup:     D->C->B->A->F->E
                   E's Backup:     E->D->C->B->A->F

                   --- NODE PROTECTION ---

                   A's Backup:     A->F->E->D->C
                   B's Backup:     B->A->F->E->D
                   C's Backup:     C->B->A->F->E
                   D's Backup:     D->C->B->A->F
                   E's Backup:     E->D->C->B->A


                         Protected and Backup LSPs

                                 Figure 5

   In case of failure, let's say on link B-C, and link protection
   configuration, the protected LSP is locally rerouted by B to C using
   the pre-computed and set-up LSP depicted in figure 5.  Considering
   that the network has a ring topology, the only other way of reaching
   C is moving counterclockwise and establishing an alternative path
   from B to C trough A, F, E, D and C. Once C has been reached, the
   traffic is re-joined to the protected LSP and delivered down to D, E
   and F.

   This is the list of nodes crossed by the traffic flow in case of
   failure on link BC:

















Ceccarelli, et al.      Expires February 19, 2010              [Page 11]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


         A-B (*)             B-A-F-E-D-C (@)                C-D-E-F (#)


   Segment "upstream"        backup path           Segment "downstream"
   with respect to                                      with respect to
   the failure                                              the failure

                                                Source
                            F                      *
                          +---+                  +*--+
                 DSLAM3<##|#  |                  |*  | A
                          |# @|@@@@@@@@@@@@@@@@@@|*@@|
                          +---+                  +---+
                           # @                    * @
                           # @                    * @
                           # @                    * @
                           # @                    * @
                          +---+                  +---+
                        E |# @|                  |@@@|B
                          |# @|                  |   |
                          +---+                  +---+
                           # @
                           # @
                           # @                   XXXXX Failure
                           # @
                          +---+                  +---+
                        D |# @|@@@@@@@@@@@@@@@@@@|#  |##>DSLAM1
                          |###|##################|#  |
                          +---+                  +---+
                             #                     C
                             #
                             V
                           DSLAM2

    FRR-TP application to ring topology - Link protection Configuration

                                 Figure 6

   This example shows that the utilization of the FRR-TP method in ring
   topology networks is not optimized in terms of bandwidth utilization
   and number of hops to be crossed.  The traffic flows through the same
   links more than once and, in this particular case, all link are used
   twice.  For further considerations concerning bandwidth utilization
   please see related paragraph.

   This recovery mechanism shows a quite big limit when providing node
   protection.  In such a case, even if a failure affects a link, the
   protection assumes that the failure affects the downstream node and



Ceccarelli, et al.      Expires February 19, 2010              [Page 12]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


   the backup path is redirected to the following node.

   It is possible to show this limit considering the previous example
   where node protection is configured.  A failure affects link BC, the
   recovery mechanism assumes that node C is failed, B redirects traffic
   towards node D (i.e. merge point) through the appropriate LSP (i.e
   [B, A, F, E, D]), then node C is excluded even if it is perfectly
   functioning and even if the network has enough resources to reach it.











































Ceccarelli, et al.      Expires February 19, 2010              [Page 13]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


         A-B (*)             B-A-F-E-D-C (@)                C-D-E-F (#)


   Segment "upstream"        backup path           Segment "downstream"
   with respect to                                      with respect to
   the failure                                              the failure

                                                Source
                            F                      *
                          +---+                  +*--+
                 DSLAM3<##|#  |                  |*  | A
                          |# @|@@@@@@@@@@@@@@@@@@|*@@|
                          +---+                  +---+
                           # @                    * @
                           # @                    * @
                           # @                    * @
                           # @                    * @
                          +---+                  +---+
                        E |# @|                  |@@@|B
                          |# @|                  |   |
                          +---+                  +---+
                           # @
                           # @
                           # @                   XXXXX Failure
                           # @
                          +---+                  +---+
                        D |# @|                  |   |-->!DSLAM1!
                          |###|------------------|   |
                          +---+                  +---+
                             #                     C
                             #
                             V
                           DSLAM2

    FRR-TP application to ring topology - Node protection Configuration

                                 Figure 7

6.2.  Optimized Multicast Fast Rerouting (ROM-FRR)

   ROM-FRR is a P2MP recovery mechanism based on FRR.  It behaves in the
   same manner as FRR-TP with the difference that it does not provide a
   backup path between the end nodes of a failed link (link protection)
   or between the upstream and downstream node of a failed node (node
   protection), but a backup P2MP LSP from the upstream node with
   respect to the failure and all the destinations downstream the
   failure.




Ceccarelli, et al.      Expires February 19, 2010              [Page 14]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


   Considering the example depicted in figure 4 it is possible to
   identify the protected LSP and all the possible backup LSPs.  "X's
   Backup" is the backup path activated by X as a consequence of a
   failure affecting node Y (downstream node with respect to X) or link
   X-Y, and square brackets indicate nodes delivering traffic to the
   DSLAMs.



                   Protected LSP:  A->B->[C]->[D]->E->[F]

                   --- LINK/NODE PROTECTION ---

                   A's Backup:     A->[F]->E->[D]->[C]
                   B's Backup:     B->A->[F]->E->[D]->[C]
                   C's Backup:     C->B->A->[F]->E->[D]
                   D's Backup:     D->C->B->A->[F]
                   E's Backup:     E->D->C->B->A->[F]


                         Protected and Backup LSPs

                                 Figure 8

   ROM-FRR can be applied to any network topology but it is particularly
   efficient for interconnected ring topologies.

   In the following it is foreseen an example comparing the application
   of the FRR-TP and ROM-FRR to the use case previously seen.  The
   topology is the same and the failure is considered to occur on link
   BC.  This is the list of nodes involved in the protection.  This time
   we highlight in brackets the nodes that drop and continue traffic
   from the ring.


















Ceccarelli, et al.      Expires February 19, 2010              [Page 15]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


   -- FRR-TP --

           A-B               B-A-F-E-D-C               [C]-[D]-E-[F]


   Segment "upstream"        backup path           Segment "downstream"
   with respect to                                      with respect to
   the failure                                              the failure

   -- ROM-FRR --

           A-B (*)           B-A-[F]-E-[D]-[C] (@)


   Segment "upstream"        backup path
   with respect to
   the failure
                                                Source
                            F                      *
                          +---+                  +*--+
                 DSLAM3<@@|@  |                  |*  | A
                          |  @|@@@@@@@@@@@@@@@@@@|*@@|
                          +---+                  +---+
                             @                    * @
                             @                    * @
                             @                    * @
                             @                    * @
                          +---+                  +---+
                        E |  @|                  |@@@|B
                          |  @|                  |   |
                          +---+                  +---+
                             @
                             @
                             @                   XXXXX Failure
                             @
                          +---+                  +---+

                          D |  @|                  | @@|@@>DSLAM1
                          |  @|@@@@@@@@@@@@@@@@@@|@  |
                          +---+                  +---+
                             @                     C
                             @
                             V
                           DSLAM2



                             FRR-TP vs ROM-FRR



Ceccarelli, et al.      Expires February 19, 2010              [Page 16]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


                                 Figure 9

   Comparing the two lists of nodes, it is possible to see that in this
   particular case the number of hops crossed using FRR-TP is
   significantly higher than the number of hops made by the traffic when
   ROM-FRR is used (generally it is always higher or at least equal).
   This implies a sensible waste of bandwidth on all those links that
   are crossed in both directions.

   Moreover FRR-TP is configured to face with a specific failure: link
   protection or node protection.  If the protection is configured to
   react to a node failure but the failure affects a link, this results
   in isolating node C so failing to feed DSLAM1.

   This problem doesn't happen in case of ROM-FRR, because there is no
   distinction between node protection and link protection.  In case of
   link BC or node C failure, the rerouting is performed in both cases
   attempting to reach C. If the failure regards the link, the traffic
   is correctly delivered to C, while if the failure affects node C, the
   rerouting is correctly performed up to node D.

6.3.  Bandwidth Utilization Comparison

   Considering the ring network previously seen, it is possible to do
   some bandwidth utilization considerations.  The protected LSP is set
   up from A to F clockwise and an X Mbps bandwidth is reserved along
   the path.  All the backup LSPs are preprovisioned counterclockwise,
   each of them with reserved bandwidth X. These LSPs share the same
   bandwidth in a SE (Shared Explicit) [RFC 2205] style.

   The bandwidth reserved counterclockwise is not used when the
   protected LSP is properly working and can be used for extra traffic
   [RFC 4427].

   In case of failure, the bandwidth actually used differs depending on
   the recovery mechanism implemented.  In case of FRR-TP, the bandwidth
   used is X on both directions of almost each link, while in case of
   ROM-FRR only the links from the ingress node to the node detecting
   the failure have an X bandwidth used in both directions, while all
   the others have an X bandwidth only in the counterclockwise
   direction.

   Let's suppose to have a failure on link B-C shown in figure 4.  In
   the following table it is possible to find the Bandwidth utilization
   on each link (always equal to X), for each recovery mechanism and for
   each direction (CW=clockwise, CCW=counterclockwise).





Ceccarelli, et al.      Expires February 19, 2010              [Page 17]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


   +----------------------------+----------------------------+
   |           FRR-TP           |           ROM-FRR          |
   +--------------+-------------+--------------+-------------+
   |   Link A-B   |    CW+CCW   |   Link A-B   |    CW+CCW   |
   |   Link A-F   |     CCW     |   Link A-F   |     CCW     |
   |   Link F-E   |    CW+CCW   |   Link F-E   |     CCW     |
   |   Link E-D   |    CW+CCW   |   Link E-D   |     CCW     |
   |   Link D-C   |    CW+CCW   |   Link D-C   |     CCW     |
   +--------------+-------------+--------------+-------------+


                     Bandwidth utilization comparison

                                 Figure 10

6.4.  Multiple Failures Comparison

   A further comparison between FRR-TP and ROM-FRR can be done with
   respect to their ability to react to multiple failures.  FRR-TP
   recovery mechanism hasn't the ability to recover from multiple
   failures on a ring network, while ROM-FRR is able to recover, from
   multiple failures.

   Let's consider, for example, a double link failure affecting links
   B-C and C-D shown in figure 4.  The FRR-TP mechanism is not able to
   recover from the failure because B, upon detecting the failure, has
   no alternative paths to reach C. The whole P2MP traffic is lost.
   ROM-FRR mechanism is able to partially recover from the failure, in
   fact the backup P2MP LSP to node F and node D is correctly set up and
   DSLAMs 2 and 3 continue receiving traffic.


7.  Conclusions

   The comparison of the FRR-TP an ROM-FRR methods leads to the
   following assumptions:

      1.  ROM-FRR allows a sensible save of bandwidth.  With respect to
      figure 7 it is possible to see that only the links from the
      ingress node to the failure (clockwise) are crossed in both ways
      (A-B), while all the other ones (from the ingress to failure,
      counterclockwise) are crossed just once(F-E-D-C).

      2.  The average number of hops crossed using FRR-TP is
      significantly higher as a consequence of the previous bullet.  In
      the example shown in figure 7, the "farthest" node is reached in 9
      hops using FRR-TP and only 6 hops using ROM-FRR.  In general the
      number of hops is always lower (or equal in the worst case) using



Ceccarelli, et al.      Expires February 19, 2010              [Page 18]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


      ROM-FRR.

      3.  FRR-TP must be configured in node protection or link
      protection mode.  This leads to a bug in case of link failure and
      node protection configured, when the node supposed to be failed is
      perfectly working but indeed isolated by the backup up.

      4.  FRR-TP is not able to react to multiple failures, while ROM-
      FRR is able to partially recover from multiple failure, depending
      on the resources affected.


8.  Acknowledgements

   TBD


9.  IANA Considerations

   This memo includes no request to IANA.


10.  Security Considerations

   This section will be added in a future version.


11.  References

11.1.  Normative References

   [RFC3031]  Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol
              Label Switching Architecture", RFC 3031, January 2001.

   [RFC4875]  Aggarwal, R., Papadimitriou, D., and S. Yasukawa,
              "Extensions to Resource Reservation Protocol - Traffic
              Engineering (RSVP-TE) for Point-to-Multipoint TE Label
              Switched Paths (LSPs)", RFC 4875, May 2007.

   [RFC4090]  Pan, P., Swallow, G., and A. Atlas, "Fast Reroute
              Extensions to RSVP-TE for LSP Tunnels", RFC 4090,
              May 2005.

   [RFC2205]  Braden, B., Zhang, L., Berson, S., Herzog, S., and S.
              Jamin, "Resource ReSerVation Protocol (RSVP) -- Version 1
              Functional Specification", RFC 2205, September 1997.

   [RFC4427]  Mannie, E. and D. Papadimitriou, "Recovery (Protection and



Ceccarelli, et al.      Expires February 19, 2010              [Page 19]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


              Restoration) Terminology for Generalized Multi-Protocol
              Label Switching (GMPLS)", RFC 4427, March 2006.

   [RFC3812]  Srinivasan, C., Viswanathan, A., and T. Nadeau,
              "Multiprotocol Label Switching (MPLS) Traffic Engineering
              (TE) Management Information Base (MIB)", RFC 3812,
              June 2004.

   [RFC3813]  Srinivasan, C., Viswanathan, A., and T. Nadeau,
              "Multiprotocol Label Switching (MPLS) Label Switching
              Router (LSR) Management Information Base (MIB)", RFC 3813,
              June 2004.

   [DRAFT-JENKINS]
              Niven-Jenkins, B., "MPLS-TP Requirements", 2008.

   [DRAFT-SPRECHER]
              Sprecher, N., "MPLS-TP OAM Analysis", 2008.

   [DRAFT-BLB]
              Bocci, M., "A Framework for MPLS in Transport Networks",
              2008.

   [DRAFT-PWE3-P2MP]
              Martini, L., "Signaling Root-Initiated Point-to-Multipoint
              Pseudowires using LDP", 2008.

   [DRAFT-P2MP-TE-MIB]
              Farrel, A., "Point-to-Multipoint Multiprotocol Label
              Switching (MPLS) Traffic Engineering (TE) Management
              Information Base (MIB) module", 2008.

11.2.  Informational References

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


Authors' Addresses

   Daniele Ceccarelli
   Ericsson
   Via A. Negrone 1/A
   Genova - Sestri Ponente
   Italy

   Email: daniele.ceccarelli@ericsson.com




Ceccarelli, et al.      Expires February 19, 2010              [Page 20]


Internet-Draft    draft-ceccarelli-mpls-tp-p2mp-ring-01      August 2009


   Diego Caviglia
   Ericsson
   Via A. Negrone 1/A
   Genova - Sestri Ponente
   Italy

   Email: diego.caviglia@ericsson.com


   Francesco Fondelli
   Ericsson
   Via A. Negrone 1/A
   Genova - Sestri Ponente
   Italy

   Email: francesco.fondelli@ericsson.com


   Marco Corsi
   Altran
   Via A. Negrone 1/A
   Genova - Sestri Ponente
   Italy

   Email: marco.corsi@altran.it


   Alessandro D'Alessandro
   Telecom Italia
   Via G. Reiss Romoli, 274
   Torino
   Italy

   Email: alessandro.dalessandro@telecomitalia.it

















Ceccarelli, et al.      Expires February 19, 2010              [Page 21]