Rfc8670
TitleBGP Prefix Segment in Large-Scale Data Centers
AuthorC. Filsfils, Ed., S. Previdi, G. Dawra, E. Aries, P. Lapukhov
DateDecember 2019
Format:HTML, TXT, PDF, XML
Status:INFORMATIONAL





Internet Engineering Task Force (IETF)                  C. Filsfils, Ed.
Request for Comments: 8670                                    S. Previdi
Category: Informational                              Cisco Systems, Inc.
ISSN: 2070-1721                                                 G. Dawra
                                                                LinkedIn
                                                                E. Aries
                                                            Arrcus, Inc.
                                                             P. Lapukhov
                                                                Facebook
                                                           December 2019


             BGP Prefix Segment in Large-Scale Data Centers

Abstract

   This document describes the motivation for, and benefits of, applying
   Segment Routing (SR) in BGP-based large-scale data centers.  It
   describes the design to deploy SR in those data centers for both the
   MPLS and IPv6 data planes.

Status of This Memo

   This document is not an Internet Standards Track specification; it is
   published for informational purposes.

   This document is a product of the Internet Engineering Task Force
   (IETF).  It represents the consensus of the IETF community.  It has
   received public review and has been approved for publication by the
   Internet Engineering Steering Group (IESG).  Not all documents
   approved by the IESG are candidates for any level of Internet
   Standard; see Section 2 of RFC 7841.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   https://www.rfc-editor.org/info/rfc8670.

Copyright Notice

   Copyright (c) 2019 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (https://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.  Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the Simplified BSD License.

Table of Contents

   1.  Introduction
   2.  Large-Scale Data-Center Network Design Summary
     2.1.  Reference Design
   3.  Some Open Problems in Large Data-Center Networks
   4.  Applying Segment Routing in the DC with MPLS Data Plane
     4.1.  BGP Prefix Segment (BGP Prefix-SID)
     4.2.  EBGP Labeled Unicast (RFC 8277)
       4.2.1.  Control Plane
       4.2.2.  Data Plane
       4.2.3.  Network Design Variation
       4.2.4.  Global BGP Prefix Segment through the Fabric
       4.2.5.  Incremental Deployments
     4.3.  IBGP Labeled Unicast (RFC 8277)
   5.  Applying Segment Routing in the DC with IPv6 Data Plane
   6.  Communicating Path Information to the Host
   7.  Additional Benefits
     7.1.  MPLS Data Plane with Operational Simplicity
     7.2.  Minimizing the FIB Table
     7.3.  Egress Peer Engineering
     7.4.  Anycast
   8.  Preferred SRGB Allocation
   9.  IANA Considerations
   10. Manageability Considerations
   11. Security Considerations
   12. References
     12.1.  Normative References
     12.2.  Informative References
   Acknowledgements
   Contributors
   Authors' Addresses

1.  Introduction

   Segment Routing (SR), as described in [RFC8402], leverages the
   source-routing paradigm.  A node steers a packet through an ordered
   list of instructions called "segments".  A segment can represent any
   instruction, topological or service based.  A segment can have a
   local semantic to an SR node or a global semantic within an SR
   domain.  SR allows the enforcement of a flow through any topological
   path while maintaining per-flow state only from the ingress node to
   the SR domain.  SR can be applied to the MPLS and IPv6 data planes.

   The use cases described in this document should be considered in the
   context of the BGP-based large-scale data-center (DC) design
   described in [RFC7938].  This document extends it by applying SR both
   with IPv6 and MPLS data planes.

2.  Large-Scale Data-Center Network Design Summary

   This section provides a brief summary of the Informational RFC
   [RFC7938], which outlines a practical network design suitable for
   data centers of various scales:

   *  Data-center networks have highly symmetric topologies with
      multiple parallel paths between two server-attachment points.  The
      well-known Clos topology is most popular among the operators (as
      described in [RFC7938]).  In a Clos topology, the minimum number
      of parallel paths between two elements is determined by the
      "width" of the "Tier-1" stage.  See Figure 1 for an illustration
      of the concept.

   *  Large-scale data centers commonly use a routing protocol, such as
      BGP-4 [RFC4271], in order to provide endpoint connectivity.
      Therefore, recovery after a network failure is driven either by
      local knowledge of directly available backup paths or by
      distributed signaling between the network devices.

   *  Within data-center networks, traffic is load shared using the
      Equal Cost Multipath (ECMP) mechanism.  With ECMP, every network
      device implements a pseudorandom decision, mapping packets to one
      of the parallel paths by means of a hash function calculated over
      certain parts of the packet, typically a combination of various
      packet header fields.

   The following is a schematic of a five-stage Clos topology with four
   devices in the "Tier-1" stage.  Notice that the number of paths
   between Node1 and Node12 equals four; the paths have to cross all of
   the Tier-1 devices.  At the same time, the number of paths between
   Node1 and Node2 equals two, and the paths only cross Tier-2 devices.
   Other topologies are possible, but for simplicity, only the
   topologies that have a single path from Tier-1 to Tier-3 are
   considered below.  The rest could be treated similarly, with a few
   modifications to the logic.

2.1.  Reference Design

                                   Tier-1
                                  +-----+
                                  |NODE |
                               +->|  5  |--+
                               |  +-----+  |
                       Tier-2  |           |   Tier-2
                      +-----+  |  +-----+  |  +-----+
        +------------>|NODE |--+->|NODE |--+--|NODE |-------------+
        |       +-----|  3  |--+  |  6  |  +--|  9  |-----+       |
        |       |     +-----+     +-----+     +-----+     |       |
        |       |                                         |       |
        |       |     +-----+     +-----+     +-----+     |       |
        | +-----+---->|NODE |--+  |NODE |  +--|NODE |-----+-----+ |
        | |     | +---|  4  |--+->|  7  |--+--|  10 |---+ |     | |
        | |     | |   +-----+  |  +-----+  |  +-----+   | |     | |
        | |     | |            |           |            | |     | |
      +-----+ +-----+          |  +-----+  |          +-----+ +-----+
      |NODE | |NODE | Tier-3   +->|NODE |--+   Tier-3 |NODE | |NODE |
      |  1  | |  2  |             |  8  |             | 11  | |  12 |
      +-----+ +-----+             +-----+             +-----+ +-----+
        | |     | |                                     | |     | |
        A O     B O            <- Servers ->            Z O     O O

                      Figure 1: 5-Stage Clos Topology

   In the reference topology illustrated in Figure 1, it is assumed:

   *  Each node is its own autonomous system (AS) (Node X has AS X).
      4-byte AS numbers are recommended ([RFC6793]).

      -  For simple and efficient route propagation filtering, Node5,
         Node6, Node7, and Node8 use the same AS; Node3 and Node4 use
         the same AS; and Node9 and Node10 use the same AS.

      -  In the case in which 2-byte autonomous system numbers are used
         for efficient usage of the scarce 2-byte Private Use AS pool,
         different Tier-3 nodes might use the same AS.

      -  Without loss of generality, these details will be simplified in
         this document.  It is to be assumed that each node has its own
         AS.

   *  Each node peers with its neighbors with a BGP session.  If not
      specified, external BGP (EBGP) is assumed.  In a specific use
      case, internal BGP (IBGP) will be used, but this will be called
      out explicitly in that case.

   *  Each node originates the IPv4 address of its loopback interface
      into BGP and announces it to its neighbors.

      -  The loopback of Node X is 192.0.2.x/32.

   In this document, the Tier-1, Tier-2, and Tier-3 nodes are referred
   to as "Spine", "Leaf", and "ToR" (top of rack) nodes, respectively.
   When a ToR node acts as a gateway to the "outside world", it is
   referred to as a "border node".

3.  Some Open Problems in Large Data-Center Networks

   The data-center-network design summarized above provides means for
   moving traffic between hosts with reasonable efficiency.  There are
   few open performance and reliability problems that arise in such a
   design:

   *  ECMP routing is most commonly realized per flow.  This means that
      large, long-lived "elephant" flows may affect performance of
      smaller, short-lived "mouse" flows and may reduce efficiency of
      per-flow load sharing.  In other words, per-flow ECMP does not
      perform efficiently when flow-lifetime distribution is heavy
      tailed.  Furthermore, due to hash-function inefficiencies, it is
      possible to have frequent flow collisions where more flows get
      placed on one path over the others.

   *  Shortest-path routing with ECMP implements an oblivious routing
      model that is not aware of the network imbalances.  If the network
      symmetry is broken, for example, due to link failures, utilization
      hotspots may appear.  For example, if a link fails between Tier-1
      and Tier-2 devices (e.g., Node5 and Node9), Tier-3 devices Node1
      and Node2 will not be aware of that since there are other paths
      available from the perspective of Node3.  They will continue
      sending roughly equal traffic to Node3 and Node4 as if the failure
      didn't exist, which may cause a traffic hotspot.

   *  Isolating faults in the network with multiple parallel paths and
      ECMP-based routing is nontrivial due to lack of determinism.
      Specifically, the connections from HostA to HostB may take a
      different path every time a new connection is formed, thus making
      consistent reproduction of a failure much more difficult.  This
      complexity scales linearly with the number of parallel paths in
      the network and stems from the random nature of path selection by
      the network devices.

4.  Applying Segment Routing in the DC with MPLS Data Plane

4.1.  BGP Prefix Segment (BGP Prefix-SID)

   A BGP Prefix Segment is a segment associated with a BGP prefix.  A
   BGP Prefix Segment is a network-wide instruction to forward the
   packet along the ECMP-aware best path to the related prefix.

   The BGP Prefix Segment is defined as the BGP Prefix-SID Attribute in
   [RFC8669], which contains an index.  Throughout this document, the
   BGP Prefix Segment Attribute is referred to as the "BGP Prefix-SID"
   and the encoded index as the label index.

   In this document, the network design decision has been made to assume
   that all the nodes are allocated the same SRGB (Segment Routing
   Global Block), e.g., [16000, 23999].  This provides operational
   simplification as explained in Section 8, but this is not a
   requirement.

   For illustration purposes, when considering an MPLS data plane, it is
   assumed that the label index allocated to prefix 192.0.2.x/32 is X.
   As a result, a local label (16000+x) is allocated for prefix
   192.0.2.x/32 by each node throughout the DC fabric.

   When the IPv6 data plane is considered, it is assumed that Node X is
   allocated IPv6 address (segment) 2001:DB8::X.

4.2.  EBGP Labeled Unicast (RFC 8277)

   Referring to Figure 1 and [RFC7938], the following design
   modifications are introduced:

   *  Each node peers with its neighbors via an EBGP session with
      extensions defined in [RFC8277] (named "EBGP8277" throughout this
      document) and with the BGP Prefix-SID attribute extension as
      defined in [RFC8669].

   *  The forwarding plane at Tier-2 and Tier-1 is MPLS.

   *  The forwarding plane at Tier-3 is either IP2MPLS (if the host
      sends IP traffic) or MPLS2MPLS (if the host sends MPLS-
      encapsulated traffic).

   Figure 2 zooms into a path from ServerA to ServerZ within the
   topology of Figure 1.

                      +-----+     +-----+     +-----+
          +---------->|NODE |     |NODE |     |NODE |
          |           |  4  |--+->|  7  |--+--|  10 |---+
          |           +-----+     +-----+     +-----+   |
          |                                             |
      +-----+                                         +-----+
      |NODE |                                         |NODE |
      |  1  |                                         | 11  |
      +-----+                                         +-----+
        |                                              |
        A                    <- Servers ->             Z

          Figure 2: Path from A to Z via Nodes 1, 4, 7, 10, and 11

   Referring to Figures 1 and 2, and assuming the IP address with the AS
   and label-index allocation previously described, the following
   sections detail the control-plane operation and the data-plane states
   for the prefix 192.0.2.11/32 (loopback of Node11).

4.2.1.  Control Plane

   Node11 originates 192.0.2.11/32 in BGP and allocates to it a BGP
   Prefix-SID with label-index: index11 [RFC8669].

   Node11 sends the following EBGP8277 update to Node10:

      IP Prefix:  192.0.2.11/32

      Label:  Implicit NULL

      Next hop:  Node11's interface address on the link to Node10

      AS Path:  {11}

      BGP Prefix-SID:  Label-Index 11

   Node10 receives the above update.  As it is SR capable, Node10 is
   able to interpret the BGP Prefix-SID; therefore, it understands that
   it should allocate the label from its own SRGB block, offset by the
   label index received in the BGP Prefix-SID (16000+11, hence, 16011)
   to the Network Layer Reachability Information (NLRI) instead of
   allocating a nondeterministic label out of a dynamically allocated
   portion of the local label space.  The implicit NULL label in the
   NLRI tells Node10 that it is the penultimate hop and that it must pop
   the top label on the stack before forwarding traffic for this prefix
   to Node11.

   Then, Node10 sends the following EBGP8277 update to Node7:

      IP Prefix:  192.0.2.11/32

      Label:  16011

      Next hop:  Node10's interface address on the link to Node7

      AS Path:  {10, 11}

      BGP Prefix-SID:  Label-Index 11

   Node7 receives the above update.  As it is SR capable, Node7 is able
   to interpret the BGP Prefix-SID; therefore, it allocates the local
   (incoming) label 16011 (16000 + 11) to the NLRI (instead of
   allocating a "dynamic" local label from its label manager).  Node7
   uses the label in the received EBGP8277 NLRI as the outgoing label
   (the index is only used to derive the local/incoming label).

   Node7 sends the following EBGP8277 update to Node4:

      IP Prefix:  192.0.2.11/32

      Label:  16011

      Next hop:  Node7's interface address on the link to Node4

      AS Path:  {7, 10, 11}

      BGP Prefix-SID:  Label-Index 11

   Node4 receives the above update.  As it is SR capable, Node4 is able
   to interpret the BGP Prefix-SID; therefore, it allocates the local
   (incoming) label 16011 to the NLRI (instead of allocating a "dynamic"
   local label from its label manager).  Node4 uses the label in the
   received EBGP8277 NLRI as an outgoing label (the index is only used
   to derive the local/incoming label).

   Node4 sends the following EBGP8277 update to Node1:

      IP Prefix:  192.0.2.11/32

      Label:  16011

      Next hop:  Node4's interface address on the link to Node1

      AS Path:  {4, 7, 10, 11}

      BGP Prefix-SID:  Label-Index 11

   Node1 receives the above update.  As it is SR capable, Node1 is able
   to interpret the BGP Prefix-SID; therefore, it allocates the local
   (incoming) label 16011 to the NLRI (instead of allocating a "dynamic"
   local label from its label manager).  Node1 uses the label in the
   received EBGP8277 NLRI as an outgoing label (the index is only used
   to derive the local/incoming label).

4.2.2.  Data Plane

   Referring to Figure 1, and assuming all nodes apply the same
   advertisement rules described above and all nodes have the same SRGB
   (16000-23999), here are the IP/MPLS forwarding tables for prefix
   192.0.2.11/32 at Node1, Node4, Node7, and Node10.

    +----------------------------------+----------------+------------+
    | Incoming Label or IP Destination | Outgoing Label |  Outgoing  |
    |                                  |                | Interface  |
    +----------------------------------+----------------+------------+
    |              16011               |     16011      | ECMP{3, 4} |
    +----------------------------------+----------------+------------+
    |          192.0.2.11/32           |     16011      | ECMP{3, 4} |
    +----------------------------------+----------------+------------+

                     Table 1: Node1 Forwarding Table

    +----------------------------------+----------------+------------+
    | Incoming Label or IP Destination | Outgoing Label |  Outgoing  |
    |                                  |                | Interface  |
    +----------------------------------+----------------+------------+
    |              16011               |     16011      | ECMP{7, 8} |
    +----------------------------------+----------------+------------+
    |          192.0.2.11/32           |     16011      | ECMP{7, 8} |
    +----------------------------------+----------------+------------+

                     Table 2: Node4 Forwarding Table

     +----------------------------------+----------------+-----------+
     | Incoming Label or IP Destination | Outgoing Label |  Outgoing |
     |                                  |                | Interface |
     +----------------------------------+----------------+-----------+
     |              16011               |     16011      |     10    |
     +----------------------------------+----------------+-----------+
     |          192.0.2.11/32           |     16011      |     10    |
     +----------------------------------+----------------+-----------+

                      Table 3: Node7 Forwarding Table

     +----------------------------------+----------------+-----------+
     | Incoming Label or IP Destination | Outgoing Label |  Outgoing |
     |                                  |                | Interface |
     +----------------------------------+----------------+-----------+
     |              16011               |      POP       |     11    |
     +----------------------------------+----------------+-----------+
     |          192.0.2.11/32           |      N/A       |     11    |
     +----------------------------------+----------------+-----------+

                      Table 4: Node10 Forwarding Table

4.2.3.  Network Design Variation

   A network design choice could consist of switching all the traffic
   through Tier-1 and Tier-2 as MPLS traffic.  In this case, one could
   filter away the IP entries at Node4, Node7, and Node10.  This might
   be beneficial in order to optimize the forwarding table size.

   A network design choice could consist of allowing the hosts to send
   MPLS-encapsulated traffic based on the Egress Peer Engineering (EPE)
   use case as defined in [SR-CENTRAL-EPE].  For example, applications
   at HostA would send their Z-destined traffic to Node1 with an MPLS
   label stack where the top label is 16011 and the next label is an EPE
   peer segment ([SR-CENTRAL-EPE]) at Node11 directing the traffic to Z.

4.2.4.  Global BGP Prefix Segment through the Fabric

   When the previous design is deployed, the operator enjoys global BGP
   Prefix-SID and label allocation throughout the DC fabric.

   A few examples follow:

   *  Normal forwarding to Node11: A packet with top label 16011
      received by any node in the fabric will be forwarded along the
      ECMP-aware BGP best path towards Node11, and the label 16011 is
      penultimate popped at Node10 (or at Node 9).

   *  Traffic-engineered path to Node11: An application on a host behind
      Node1 might want to restrict its traffic to paths via the Spine
      node Node5.  The application achieves this by sending its packets
      with a label stack of {16005, 16011}. BGP Prefix-SID 16005 directs
      the packet up to Node5 along the path (Node1, Node3, Node5).  BGP
      Prefix-SID 16011 then directs the packet down to Node11 along the
      path (Node5, Node9, Node11).

4.2.5.  Incremental Deployments

   The design previously described can be deployed incrementally.  Let
   us assume that Node7 does not support the BGP Prefix-SID, and let us
   show how the fabric connectivity is preserved.

   From a signaling viewpoint, nothing would change; even though Node7
   does not support the BGP Prefix-SID, it does propagate the attribute
   unmodified to its neighbors.

   From a label-allocation viewpoint, the only difference is that Node7
   would allocate a dynamic (random) label to the prefix 192.0.2.11/32
   (e.g., 123456) instead of the "hinted" label as instructed by the BGP
   Prefix-SID.  The neighbors of Node7 adapt automatically as they
   always use the label in the BGP8277 NLRI as an outgoing label.

   Node4 does understand the BGP Prefix-SID; therefore, it allocates the
   indexed label in the SRGB (16011) for 192.0.2.11/32.

   As a result, all the data-plane entries across the network would be
   unchanged except the entries at Node7 and its neighbor Node4 as shown
   in the figures below.

   The key point is that the end-to-end Label Switched Path (LSP) is
   preserved because the outgoing label is always derived from the
   received label within the BGP8277 NLRI.  The index in the BGP Prefix-
   SID is only used as a hint on how to allocate the local label (the
   incoming label) but never for the outgoing label.

     +----------------------------------+----------------+-----------+
     | Incoming Label or IP Destination | Outgoing Label |  Outgoing |
     |                                  |                | Interface |
     +----------------------------------+----------------+-----------+
     |              12345               |     16011      |     10    |
     +----------------------------------+----------------+-----------+

                      Table 5: Node7 Forwarding Table

     +----------------------------------+----------------+-----------+
     | Incoming Label or IP Destination | Outgoing Label |  Outgoing |
     |                                  |                | Interface |
     +----------------------------------+----------------+-----------+
     |              16011               |     12345      |     7     |
     +----------------------------------+----------------+-----------+

                      Table 6: Node4 Forwarding Table

   The BGP Prefix-SID can thus be deployed incrementally, i.e., one node
   at a time.

   When deployed together with a homogeneous SRGB (the same SRGB across
   the fabric), the operator incrementally enjoys the global prefix
   segment benefits as the deployment progresses through the fabric.

4.3.  IBGP Labeled Unicast (RFC 8277)

   The same exact design as EBGP8277 is used with the following
   modifications:

   *  All nodes use the same AS number.

   *  Each node peers with its neighbors via an internal BGP session
      (IBGP) with extensions defined in [RFC8277] (named "IBGP8277"
      throughout this document).

   *  Each node acts as a route reflector for each of its neighbors and
      with the next-hop-self option.  Next-hop-self is a well-known
      operational feature that consists of rewriting the next hop of a
      BGP update prior to sending it to the neighbor.  Usually, it's a
      common practice to apply next-hop-self behavior towards IBGP peers
      for EBGP-learned routes.  In the case outlined in this section, it
      is proposed to use the next-hop-self mechanism also to IBGP-
      learned routes.

                                  Cluster-1
                               +-----------+
                               |  Tier-1   |
                               |  +-----+  |
                               |  |NODE |  |
                               |  |  5  |  |
                    Cluster-2  |  +-----+  |  Cluster-3
                   +---------+ |           | +---------+
                   | Tier-2  | |           | |  Tier-2 |
                   | +-----+ | |  +-----+  | | +-----+ |
                   | |NODE | | |  |NODE |  | | |NODE | |
                   | |  3  | | |  |  6  |  | | |  9  | |
                   | +-----+ | |  +-----+  | | +-----+ |
                   |         | |           | |         |
                   |         | |           | |         |
                   | +-----+ | |  +-----+  | | +-----+ |
                   | |NODE | | |  |NODE |  | | |NODE | |
                   | |  4  | | |  |  7  |  | | |  10 | |
                   | +-----+ | |  +-----+  | | +-----+ |
                   +---------+ |           | +---------+
                               |           |
                               |  +-----+  |
                               |  |NODE |  |
             Tier-3            |  |  8  |  |         Tier-3
         +-----+ +-----+       |  +-----+  |      +-----+ +-----+
         |NODE | |NODE |       +-----------+      |NODE | |NODE |
         |  1  | |  2  |                          | 11  | |  12 |
         +-----+ +-----+                          +-----+ +-----+

         Figure 3: IBGP Sessions with Reflection and Next-Hop-Self

   *  For simple and efficient route propagation filtering and as
      illustrated in Figure 3:

      -  Node5, Node6, Node7, and Node8 use the same Cluster ID
         (Cluster-1).

      -  Node3 and Node4 use the same Cluster ID (Cluster-2).

      -  Node9 and Node10 use the same Cluster ID (Cluster-3).

   *  The control-plane behavior is mostly the same as described in the
      previous section; the only difference is that the EBGP8277 path
      propagation is simply replaced by an IBGP8277 path reflection with
      next hop changed to self.

   *  The data-plane tables are exactly the same.

5.  Applying Segment Routing in the DC with IPv6 Data Plane

   The design described in [RFC7938] is reused with one single
   modification.  It is highlighted using the example of the
   reachability to Node11 via Spine node Node5.

   Node5 originates 2001:DB8::5/128 with the attached BGP Prefix-SID for
   IPv6 packets destined to segment 2001:DB8::5 ([RFC8402]).

   Node11 originates 2001:DB8::11/128 with the attached BGP Prefix-SID
   advertising the support of the Segment Routing Header (SRH) for IPv6
   packets destined to segment 2001:DB8::11.

   The control-plane and data-plane processing of all the other nodes in
   the fabric is unchanged.  Specifically, the routes to 2001:DB8::5 and
   2001:DB8::11 are installed in the FIB along the EBGP best path to
   Node5 (Spine node) and Node11 (ToR node) respectively.

   An application on HostA that needs to send traffic to HostZ via only
   Node5 (Spine node) can do so by sending IPv6 packets with a Segment
   Routing Header (SRH, [IPv6-SRH]).  The destination address and active
   segment is set to 2001:DB8::5.  The next and last segment is set to
   2001:DB8::11.

   The application must only use IPv6 addresses that have been
   advertised as capable for SRv6 segment processing (e.g., for which
   the BGP Prefix Segment capability has been advertised).  How
   applications learn this (e.g., centralized controller and
   orchestration) is outside the scope of this document.

6.  Communicating Path Information to the Host

   There are two general methods for communicating path information to
   the end-hosts: "proactive" and "reactive", aka "push" and "pull"
   models.  There are multiple ways to implement either of these
   methods.  Here, it is noted that one way could be using a centralized
   controller: the controller either tells the hosts of the prefix-to-
   path mappings beforehand and updates them as needed (network event
   driven push) or responds to the hosts making requests for a path to a
   specific destination (host event driven pull).  It is also possible
   to use a hybrid model, i.e., pushing some state from the controller
   in response to particular network events, while the host pulls other
   state on demand.

   Note also that when disseminating network-related data to the end-
   hosts, a trade-off is made to balance the amount of information vs.
   the level of visibility in the network state.  This applies to both
   push and pull models.  In the extreme case, the host would request
   path information on every flow and keep no local state at all.  On
   the other end of the spectrum, information for every prefix in the
   network along with available paths could be pushed and continuously
   updated on all hosts.

7.  Additional Benefits

7.1.  MPLS Data Plane with Operational Simplicity

   As required by [RFC7938], no new signaling protocol is introduced.
   The BGP Prefix-SID is a lightweight extension to BGP Labeled Unicast
   [RFC8277].  It applies either to EBGP- or IBGP-based designs.

   Specifically, LDP and RSVP-TE are not used.  These protocols would
   drastically impact the operational complexity of the data center and
   would not scale.  This is in line with the requirements expressed in
   [RFC7938].

   Provided the same SRGB is configured on all nodes, all nodes use the
   same MPLS label for a given IP prefix.  This is simpler from an
   operation standpoint, as discussed in Section 8.

7.2.  Minimizing the FIB Table

   The designer may decide to switch all the traffic at Tier-1 and
   Tier-2 based on MPLS, thereby drastically decreasing the IP table
   size at these nodes.

   This is easily accomplished by encapsulating the traffic either
   directly at the host or at the source ToR node.  The encapsulation is
   done by pushing the BGP Prefix-SID of the destination ToR for intra-
   DC traffic, or by pushing the BGP Prefix-SID for the border node for
   inter-DC or DC-to-outside-world traffic.

7.3.  Egress Peer Engineering

   It is straightforward to combine the design illustrated in this
   document with the Egress Peer Engineering (EPE) use case described in
   [SR-CENTRAL-EPE].

   In such a case, the operator is able to engineer its outbound traffic
   on a per-host-flow basis, without incurring any additional state at
   intermediate points in the DC fabric.

   For example, the controller only needs to inject a per-flow state on
   the HostA to force it to send its traffic destined to a specific
   Internet destination D via a selected border node (say Node12 in
   Figure 1 instead of another border node, Node11) and a specific
   egress peer of Node12 (say peer AS 9999 of local PeerNode segment
   9999 at Node12 instead of any other peer that provides a path to the
   destination D).  Any packet matching this state at HostA would be
   encapsulated with SR segment list (label stack) {16012, 9999}.  16012
   would steer the flow through the DC fabric, leveraging any ECMP,
   along the best path to border node Node12.  Once the flow gets to
   border node Node12, the active segment is 9999 (because of
   Penultimate Hop Popping (PHP) on the upstream neighbor of Node12).
   This EPE PeerNode segment forces border node Node12 to forward the
   packet to peer AS 9999 without any IP lookup at the border node.
   There is no per-flow state for this engineered flow in the DC fabric.
   A benefit of SR is that the per-flow state is only required at the
   source.

   As well as allowing full traffic-engineering control, such a design
   also offers FIB table-minimization benefits as the Internet-scale FIB
   at border node Node12 is not required if all FIB lookups are avoided
   there by using EPE.

7.4.  Anycast

   The design presented in this document preserves the availability and
   load-balancing properties of the base design presented in [RFC8402].

   For example, one could assign an anycast loopback 192.0.2.20/32 and
   associate segment index 20 to it on the border nodes Node11 and
   Node12 (in addition to their node-specific loopbacks).  Doing so, the
   EPE controller could express a default "go-to-the-Internet via any
   border node" policy as segment list {16020}. Indeed, from any host in
   the DC fabric or from any ToR node, 16020 steers the packet towards
   the border nodes Node11 or Node12 leveraging ECMP where available
   along the best paths to these nodes.

8.  Preferred SRGB Allocation

   In the MPLS case, it is recommended to use the same SRGBs at each
   node.

   Different SRGBs in each node likely increase the complexity of the
   solution both from an operational viewpoint and from a controller
   viewpoint.

   From an operational viewpoint, it is much simpler to have the same
   global label at every node for the same destination (the MPLS
   troubleshooting is then similar to the IPv6 troubleshooting where
   this global property is a given).

   From a controller viewpoint, this allows us to construct simple
   policies applicable across the fabric.

   Let us consider two applications, A and B, respectively connected to
   Node1 and Node2 (ToR nodes).  Application A has two flows, FA1 and
   FA2, destined to Z.  B has two flows, FB1 and FB2, destined to Z.
   The controller wants FA1 and FB1 to be load shared across the fabric
   while FA2 and FB2 must be respectively steered via Node5 and Node8.

   Assuming a consistent unique SRGB across the fabric as described in
   this document, the controller can simply do it by instructing A and B
   to use {16011} respectively for FA1 and FB1 and by instructing A and
   B to use {16005 16011} and {16008 16011} respectively for FA2 and
   FB2.

   Let us assume a design where the SRGB is different at every node and
   where the SRGB of each node is advertised using the Originator SRGB
   TLV of the BGP Prefix-SID as defined in [RFC8669]: SRGB of Node K
   starts at value K*1000, and the SRGB length is 1000 (e.g., Node1's
   SRGB is [1000, 1999], Node2's SRGB is [2000, 2999], ...).

   In this case, the controller would need to collect and store all of
   these different SRGBs (e.g., through the Originator SRGB TLV of the
   BGP Prefix-SID); furthermore, it would also need to adapt the policy
   for each host.  Indeed, the controller would instruct A to use {1011}
   for FA1 while it would have to instruct B to use {2011} for FB1
   (while with the same SRGB, both policies are the same {16011}).

   Even worse, the controller would instruct A to use {1005, 5011} for
   FA1 while it would instruct B to use {2011, 8011} for FB1 (while with
   the same SRGB, the second segment is the same across both policies:
   16011).  When combining segments to create a policy, one needs to
   carefully update the label of each segment.  This is obviously more
   error prone, more complex, and more difficult to troubleshoot.

9.  IANA Considerations

   This document has no IANA actions.

10.  Manageability Considerations

   The design and deployment guidelines described in this document are
   based on the network design described in [RFC7938].

   The deployment model assumed in this document is based on a single
   domain where the interconnected DCs are part of the same
   administrative domain (which, of course, is split into different
   autonomous systems).  The operator has full control of the whole
   domain, and the usual operational and management mechanisms and
   procedures are used in order to prevent any information related to
   internal prefixes and topology to be leaked outside the domain.

   As recommended in [RFC8402], the same SRGB should be allocated in all
   nodes in order to facilitate the design, deployment, and operations
   of the domain.

   When EPE ([SR-CENTRAL-EPE]) is used (as explained in Section 7.3),
   the same operational model is assumed.  EPE information is originated
   and propagated throughout the domain towards an internal server, and
   unless explicitly configured by the operator, no EPE information is
   leaked outside the domain boundaries.

11.  Security Considerations

   This document proposes to apply SR to a well-known scalability
   requirement expressed in [RFC7938] using the BGP Prefix-SID as
   defined in [RFC8669].

   It has to be noted, as described in Section 10, that the design
   illustrated in [RFC7938] and in this document refer to a deployment
   model where all nodes are under the same administration.  In this
   context, it is assumed that the operator doesn't want to leak outside
   of the domain any information related to internal prefixes and
   topology.  The internal information includes Prefix-SID and EPE
   information.  In order to prevent such leaking, the standard BGP
   mechanisms (filters) are applied on the boundary of the domain.

   Therefore, the solution proposed in this document does not introduce
   any additional security concerns from what is expressed in [RFC7938]
   and [RFC8669].  It is assumed that the security and confidentiality
   of the prefix and topology information is preserved by outbound
   filters at each peering point of the domain as described in
   Section 10.

12.  References

12.1.  Normative References

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

   [RFC7938]  Lapukhov, P., Premji, A., and J. Mitchell, Ed., "Use of
              BGP for Routing in Large-Scale Data Centers", RFC 7938,
              DOI 10.17487/RFC7938, August 2016,
              <https://www.rfc-editor.org/info/rfc7938>.

   [RFC8277]  Rosen, E., "Using BGP to Bind MPLS Labels to Address
              Prefixes", RFC 8277, DOI 10.17487/RFC8277, October 2017,
              <https://www.rfc-editor.org/info/rfc8277>.

   [RFC8402]  Filsfils, C., Ed., Previdi, S., Ed., Ginsberg, L.,
              Decraene, B., Litkowski, S., and R. Shakir, "Segment
              Routing Architecture", RFC 8402, DOI 10.17487/RFC8402,
              July 2018, <https://www.rfc-editor.org/info/rfc8402>.

   [RFC8669]  Previdi, S., Filsfils, C., Lindem, A., Ed., Sreekantiah,
              A., and H. Gredler, "Segment Routing Prefix Segment
              Identifier Extensions for BGP", RFC 8669,
              DOI 10.17487/RFC8669, December 2019,
              <https://www.rfc-editor.org/info/rfc8669>.

12.2.  Informative References

   [IPv6-SRH] Filsfils, C., Dukes, D., Previdi, S., Leddy, J.,
              Matsushima, S., and D. Voyer, "IPv6 Segment Routing Header
              (SRH)", Work in Progress, Internet-Draft, draft-ietf-6man-
              segment-routing-header-26, 22 October 2019,
              <https://tools.ietf.org/html/draft-ietf-6man-segment-
              routing-header-26>.

   [RFC6793]  Vohra, Q. and E. Chen, "BGP Support for Four-Octet
              Autonomous System (AS) Number Space", RFC 6793,
              DOI 10.17487/RFC6793, December 2012,
              <https://www.rfc-editor.org/info/rfc6793>.

   [SR-CENTRAL-EPE]
              Filsfils, C., Previdi, S., Dawra, G., Aries, E., and D.
              Afanasiev, "Segment Routing Centralized BGP Egress Peer
              Engineering", Work in Progress, Internet-Draft, draft-
              ietf-spring-segment-routing-central-epe-10, 21 December
              2017, <https://tools.ietf.org/html/draft-ietf-spring-
              segment-routing-central-epe-10>.

Acknowledgements

   The authors would like to thank Benjamin Black, Arjun Sreekantiah,
   Keyur Patel, Acee Lindem, and Anoop Ghanwani for their comments and
   review of this document.

Contributors

   Gaya Nagarajan
   Facebook
   United States of America

   Email: gaya@fb.com

   Gaurav Dawra
   Cisco Systems
   United States of America

   Email: gdawra.ietf@gmail.com

   Dmitry Afanasiev
   Yandex
   Russian Federation

   Email: fl0w@yandex-team.ru

   Tim Laberge
   Cisco
   United States of America

   Email: tlaberge@cisco.com

   Edet Nkposong
   Salesforce.com Inc.
   United States of America

   Email: enkposong@salesforce.com

   Mohan Nanduri
   Microsoft
   United States of America

   Email: mohan.nanduri@oracle.com

   James Uttaro
   ATT
   United States of America

   Email: ju1738@att.com

   Saikat Ray
   Unaffiliated
   United States of America

   Email: raysaikat@gmail.com

   Jon Mitchell
   Unaffiliated
   United States of America

   Email: jrmitche@puck.nether.net

Authors' Addresses

   Clarence Filsfils (editor)
   Cisco Systems, Inc.
   Brussels
   Belgium

   Email: cfilsfil@cisco.com


   Stefano Previdi
   Cisco Systems, Inc.
   Italy

   Email: stefano@previdi.net


   Gaurav Dawra
   LinkedIn
   United States of America

   Email: gdawra.ietf@gmail.com


   Ebben Aries
   Arrcus, Inc.
   2077 Gateway Place, Suite #400
   San Jose,  CA 95119
   United States of America

   Email: exa@arrcus.com


   Petr Lapukhov
   Facebook
   United States of America