Browse Prior Art Database

BIT INDEX EXPLICIT REPLICATION BEST PATH SELECTION IN LEAF-SPINE NETWORKS

IP.com Disclosure Number: IPCOM000244336D
Publication Date: 2015-Dec-02
Document File: 5 page(s) / 247K

Publishing Venue

The IP.com Prior Art Database

Related People

Neale Ranns: AUTHOR [+3]

Abstract

Mechanisms and systems are provided to select a best path from a plurality of Equal-Cost Multi Paths (ECMPs) available for forwarding multicast data packets in a leaf-spine data center network that utilizes a Bit Index Explicit Replication (BIER) architecture. To achieve optimal multicast data packet forwarding in such a leaf-spine data center network, a measure of a leaf reachability scale is incorporated into a best-path cost calculation that determines ECMPs such that in the event of a link failure traffic flow is rearranged to ensure that an ingress leaf router sends a multicast data packet only to a single spine router which replicates the multicast data packet and sends it to each leaf router in the domain.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 41% of the total text.

Page 01 of 5

BIT INDEX EXPLICIT REPLICATION BEST PATH SELECTION IN LEAF-SPINE NETWORKS

AUTHORS:

  Neale Ranns
Ijsbrand Wijnands
Greg Shepherd

CISCO SYSTEMS, INC.

ABSTRACT

    Mechanisms and systems are provided to select a best path from a plurality of Equal-Cost Multi Paths (ECMPs) available for forwarding multicast data packets in a leaf-spine data center network that utilizes a Bit Index Explicit Replication (BIER) architecture. To achieve optimal multicast data packet forwarding in such a leaf-spine data center network, a measure of a leaf reachability scale is incorporated into a best-path cost calculation that determines ECMPs such that in the event of a link failure traffic flow is rearranged to ensure that an ingress leaf router sends a multicast data packet only to a single spine router which replicates the multicast data packet and sends it to each leaf router in the domain.

DETAILED DESCRIPTION

    Leaf-spine data center networks are well known and well used network topologies. These network topologies provide, among many other benefits, a leaf router with ECMPs to each other leaf router via a single spine router. Leaf-spine data center networks may use a BIER architecture for forwarding the multicast data packets which does not require a protocol for explicitly building multicast distribution trees, nor does the BIER architecture require intermediate nodes to maintain any per-flow state. When a multicast data packet enters the domain, the ingress router determines a set of egress routers to which the packet needs to be sent. The ingress router then encapsulates the packet in a BIER header. The BIER header contains a bit string in which each bit represents exactly one egress router in the domain. To forward the packet to a given set of egress routers, the bits corresponding to those routers are set in the BIER header.

Copyright 2015 Cisco Systems, Inc.

1


Page 02 of 5

    FIG. 1 illustrates a leaf-spine data center network before a link failure occurs, where the network includes leaf routers A-E, spine routers X-Z, and ECMPs for transmitting multicast data packets.

    Two general approaches for multicast data packet delivery in a leaf-spine data center network exist. According to a first approach, the ingress leaf sends a copy of the multicast data packet to each leaf via a single spine router. For example, as shown in FIG. 1, if leaf router A wants to send a multicast data packet to leaf routers B-E, leaf router A sends only one multicast data packet to spine router X and spine router X forwards a copy of the multicast data packet to leaf routers B-E via ECMPs indicated by arrows in FIG. 1.

FIG. 1

    According to a second approach not shown in FIG. 1, an ingress leaf router (such as leaf router A in FIG. 1) may send a copy of the multicast data packet to each leaf router via a different spine router, i.e., leaf router A may send a copy of the multicast data packet to spine router X, another copy of the multicast data packet to spine router Y, and yet an...