Browse Prior Art Database

HIGHLY SCALABLE VIRTUAL EXTENSIBLE LOCAL AREA NETWORK (VXLAN) TUNNEL END POINT WITH TRAFFIC STEERING CAPABILITY IN VXLAN OVERLAYS USING PATH-BASED VIRTUAL NETWORK IDENTIFIERS

IP.com Disclosure Number: IPCOM000250449D
Publication Date: 2017-Jul-18
Document File: 6 page(s) / 270K

Publishing Venue

The IP.com Prior Art Database

Related People

Rajesh Sharma: AUTHOR [+3]

Abstract

Various path-based Virtual Network Identifiers (VNI) allocations are presented herein. In a pure decapsulation case (e.g., Virtual Extensible Local Area Network (VXLAN) to Internet Protocol (IP) address), path-based VNI allocation is made on a per logical outgoing interface basis. In the case of VXLAN to VXLAN (Data Center Interconnect (DCI) gateway), path-based VNI allocation is per {inter-DC VXLAN tunnel, tenant}. In the case of VXLAN to VXLAN (DCI gateway) with downstream VNIs, path-based VNI allocation is a function of {inter-DC VXLAN tunnel, downstream VNI (Layer 2 (L2) or Layer 3 (L3))}. For all such modes, the key benefits include skipping the inner packet lookup. Hence, routes are not needed in an Application Specific Integrated Circuit (ASIC) data plane. Also provided is the capability to pin traffic to specific paths based on user-defined policies.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 28% of the total text.

Copyright 2017 Cisco Systems, Inc. 1

HIGHLY SCALABLE VIRTUAL EXTENSIBLE LOCAL AREA NETWORK (VXLAN) TUNNEL END POINT WITH TRAFFIC STEERING CAPABILITY IN

VXLAN OVERLAYS USING PATH-BASED VIRTUAL NETWORK IDENTIFIERS

AUTHORS: Rajesh Sharma

Swaraj Kumar Chikyala Murali Garimella

CISCO SYSTEMS, INC.

ABSTRACT

Various path-based Virtual Network Identifiers (VNI) allocations are presented

herein. In a pure decapsulation case (e.g., Virtual Extensible Local Area Network (VXLAN)

to Internet Protocol (IP) address), path-based VNI allocation is made on a per logical

outgoing interface basis. In the case of VXLAN to VXLAN (Data Center Interconnect

(DCI) gateway), path-based VNI allocation is per {inter-DC VXLAN tunnel, tenant}. In

the case of VXLAN to VXLAN (DCI gateway) with downstream VNIs, path-based VNI

allocation is a function of {inter-DC VXLAN tunnel, downstream VNI (Layer 2 (L2) or

Layer 3 (L3))}. For all such modes, the key benefits include skipping the inner packet

lookup. Hence, routes are not needed in an Application Specific Integrated Circuit (ASIC)

data plane. Also provided is the capability to pin traffic to specific paths based on user-

defined policies.

DETAILED DESCRIPTION

In Virtual Extensible Local Area Network (VXLAN) – Ethernet Virtual Private

Network (EVPN) overlays, Virtual Network Identifiers (VNIs) define the lookup context

of either a bridging domain or a routing domain. In other words, when hosts are advertised

in Border Gateway Protocol (BGP) Network Layer Reachability Information (NLRIs), the

hosts are advertised with a bridging domain identifier (i.e., a Layer 2 (L2) VNI) and a

routing domain (or Virtual Routing and Forwarding (VRF)) identifier (i.e., a Layer 3 (L3)

VNI). Based on the VNI encoded in the packet, when the VXLAN encapsulated packet

arrives at the downstream VXLAN Tunnel End Point (VTEP), the forwarding lookup

context is derived and the packet match is triggered to forward the packet.

Copyright 2017 Cisco Systems, Inc. 2

The conventional forwarding model of VXLAN-EVPN enables overlay tunneling

with either bridging or routing between hosts attached to different VTEPs in the fabric.

However, this forwarding model has at least two major shortcomings. First, every subnet

or host that is announced in BGP EVPN Route-Type 5 or Route-Type 2 needs to be

installed on both upstream and downstream VTEPs. (Upstream-downstream nomenclature

is used to follow the traffic path). If the downstream VTEP is a border leaf, it can become

bottlenecked as it may have a superset of all subnet routes learned from external

connectivity and installed in hardware. Second, this forwarding model lacks the capability

to pin the traffic to specific paths once the VXLAN encapsulated traffic reaches the

downstream VTEP, decapsulated and forwarded purely based on packet match. If there are

multiple egress paths after packet lookup results, a regular hashing of packet parameters

can select one path after VXLAN decapsulation.

Described are techniques...