Scaling interconnected IP fabric data centers

Patent metadata

Field	Value
Publication number	US-9325636-B2
Application number	US-201313918658-A
Country	US
Kind code	B2
Filing date	Jun 14, 2013
Priority date	Jun 14, 2013
Publication date	Apr 26, 2016
Grant date	Apr 26, 2016

Abstract

Techniques which provide scalable techniques for managing multicast traffic in interconnected IP fabric data centers. More specifically, embodiments presented herein disclose an aggregated source technique used to address scalability issues for interconnected IP fabric data centers as well as disclose a secondary rendezvous point technique used to address backbone network (S, G) multicast state scalability. Additionally, embodiments disclosed herein include an approach for border leaf load balancing based on group destination addresses used by VTEPs.

First claim

What is claimed is:

1. A networking device in a first data center fabric connected to a second data center fabric over a backbone network to form interconnected data center fabrics configured with improved scalability in managing states, each data center fabric having a respectively plurality of endpoints of which a subset is designated as one or more border leaves of the respective data center fabric, the networking device comprising:
a computer processor; and
a memory storing an application which, when executed on the computer processor, performs an operation comprising:
receiving a packet with a header encapsulating a media access control (MAC) frame, wherein the packet has a group address as a destination address;
applying aggregated source replacement such that a total count of distinct source addresses stored in unicast states maintained in the second data center fabric scales with border leaves rather than endpoints, thereby improving scalability of maintaining unicast states in the second data center fabric, wherein applying aggregated source replacement includes replacing, in a source address field in the header, an address of an endpoint in the first data center fabric with an address of a border leaf of the first data center fabric and comprising the networking device; and
forwarding the packet towards the destination identified by the destination address;
wherein the backbone network includes a primary rendezvous point for multicast traffic spanning the first and second data center fabrics, wherein the networking device serves as a secondary rendezvous point such that a total count of multicast states maintained in the backbone network scales with border leaves rather than with endpoints functioning as multicast sources, thereby improving scalability of maintaining multicast states in the backbone network.

2. The networking device of claim 1 , wherein the first and second data center fabrics comprise first and second Internet Protocol (IP) data center fabrics, wherein the backbone network comprises an IP backbone network, wherein the border leaf connects the first IP data center fabric to the second IP data center fabric over the IP backbone network.

3. The networking device of claim 2 , wherein the operation further comprises sending one or more join messages to the source address identified in the packet to establish a multicast state in the first IP data center fabric.

4. The networking device of claim 2 , wherein the operation further comprises, prior to replacing the source address in the header, forwarding the packet towards one or more leaf networking devices in the first IP data center fabric.

5. The networking device of claim 2 , wherein the networking device is a root of a multicast tree for the first IP data center fabric.

6. The networking device of claim 1 , wherein the operation further comprises storing, in the memory, a binding between a source MAC address in the MAC frame and the source address in the header.

7. The networking device of claim 1 , wherein the operation further comprises:
receiving a second packet, wherein the second packet includes a header encapsulating a second MAC frame, wherein the header includes, as a destination address, the address of the networking device;
identifying, via the binding, a replacement destination address for the second packet;
replacing the destination address of the second packet with the replacement address; and
forwarding the second packet towards the destination identified by the destination address.

8. A computer-implemented method of improved scalability in managing states in interconnected data center fabrics including a first data center fabric connected to a second data center fabric over a backbone network, each data center fabric having a respectively plurality of endpoints of which a subset is designated as one or more border leaves of the respective data center fabric, the computer-implemented method comprising:
receiving a packet with a header encapsulating a media access control (MAC) frame, wherein the packet has a group address as a destination address;
applying aggregated source replacement such that a total count of distinct source addresses stored in unicast states maintained in the second data center fabric scales with border leaves rather than with endpoints, thereby improving scalability of maintaining unicast states in the second data center fabric, wherein applying aggregated source replacement includes replacing, in a source address field in the header, an address of an endpoint in the first data center fabric with an address of a border leaf of the first data center fabric and comprising a networking device; and
forwarding the packet towards the destination identified by the destination address;
wherein the backbone network includes a primary rendezvous point for multicast traffic spanning the first and second data center fabrics, wherein the networking device serves as a secondary rendezvous point such that a total count of multicast states maintained in the backbone network scales with border leaves rather than with endpoints functioning as multicast sources, thereby improving scalability of maintaining multicast states in the backbone network.

9. The computer-implemented method of claim 8 , wherein the first and second data center fabrics comprise first and second Internet Protocol (IP) data center fabrics, wherein the backbone network comprises an IP backbone network, wherein the border leaf connects the first IP data center fabric to the second IP data center fabric over the IP backbone network.

10. The computer-implemented method of claim 9 , further comprising sending one or more join messages to the source address identified in the packet to establish a multicast state in the first IP data center fabric.

11. The computer-implemented method of claim 9 , further comprising, prior to replacing the source address in the header, forwarding the packet towards one or more leaf networking devices in the first IP data center fabric.

12. The computer-implemented method of claim 9 , wherein the networking device is a root of a multicast tree for the first IP data center fabric.

13. The computer-implemented method of claim 8 , further comprising, storing, in the memory, a binding between a source MAC address in the MAC frame and the source address in the header.

14. The computer-implemented method of claim 8 , further comprising:
receiving a second packet, wherein the second packet includes a header encapsulating a second MAC frame, wherein the header includes, as a destination address, the address of the networking device;
identifying, via the binding, a replacement destination address for the second packet;
replacing the destination address of the second packet with the replacement address; and
forwarding the second packet towards the destination identified by the destination address.

15. A non-transitory computer-readable medium storing instructions which, when executed, performs an operation to improve scalability in managing states in interconnected data center fabrics including a first data center fabric connected to a second data center fabric over a backbone network, each data center fabric having a respectively plurality of endpoints of which a subset is designated as one or more border leaves of the respective data center fabric, the operation comprising:
receiving a packet with a header encapsulating a media access control (MAC) frame, wherein the packet has a group address as a destination address;
applying aggregated source replacement such that a total count of distinct source addresses stored in unic

Assignees

• Cisco Tech Inc

Inventors

• Yang Yibin
• Tsai Alex
• Dong Liqin
• Jin Di

Classifications

• H04L12/1886 Primary

  with traffic restrictions for efficiency improvement, e.g. involving subnets or subdomains · CPC title

• H04L12/18

  for broadcast or conference {, e.g. multicast} · CPC title

• H04L49/25 Primary

  Routing or path finding in a switch fabric · CPC title