Browse Prior Art Database

INSTANTANEOUS FLUSH OF MAC ADDRESS TABLE FOR MANY VLANS

IP.com Disclosure Number: IPCOM000223661D
Publication Date: 2012-Nov-21
Document File: 4 page(s) / 97K

Publishing Venue

The IP.com Prior Art Database

Related People

Luh-Luh Ting: AUTHOR [+3]

Abstract

A level of indirection and logic is provided to group virtual local area networks (VLANs) together so that a topology change results in fast flushing of the media access control (MAC) address tables of those VLANs. A flush can be performed with a single write to the switch ASIC MAC Lookup Table. This lightens the load on the CPU of the switch, freeing it to perform other tasks.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 48% of the total text.

Page 01 of 4

INSTANTANEOUS FLUSH OF MAC ADDRESS TABLE FOR MANY VLANS

AUTHORS:

Luh-Luh Ting

Linda Cheng

Da-Yuan Tung

CISCO SYSTEMS, INC.

ABSTRACT

    A level of indirection and logic is provided to group virtual local area networks (VLANs) together so that a topology change results in fast flushing of the media access control (MAC) address tables of those VLANs. A flush can be performed with a single write to the switch ASIC MAC Lookup Table. This lightens the load on the CPU of the switch, freeing it to perform other tasks.

DETAILED DESCRIPTION

    In existing network devices, software is able to flush the Media Access Control (MAC) Address Table for any virtual local area network (VLAN) with a single write to the VLAN's epoch field. When it increments the epoch, it causes all the Layer 2 (L2) entries in the MAC lookup table to mismatch, which in turn causes the L2 source (src) lookups to be relearned and L2 destination lookups to flood. When entries are learned on the new topology they are populated into the MAC lookup table with the incremented epoch. The use of epochs helps convergence time when topology changes on a VLAN. In particular, this feature helps convergence time on Resilient Ethernet Protocol (REP) and STP, loop detection protocols. Because REP supports VLAN load balancing, when a link fails, only the VLANs supported on that link need flooding, flushing, and relearning on that ring. Customers have complained that convergence is slow when using VLAN load balancing.

    One solution is a write to a VLAN's epoch, which takes care of flushing all the entries for that one VLAN. However, software still needs to update the epoch for each

Copyright 2012 Cisco Systems, Inc. 1


Page 02 of 4

VLAN affected, which could be in the thousands. A proposed solution is to add an extension to the epoch field, called the vlanGroup. It provides a level of indirection that allows the flushing of sets of VLANs. Consider the network drawn in Figure 1 below. It is composed of two interconnecting REP segments, red and blue. When VLAN load balancing is enabled on switch SW1, it splits the traffic in two directions. The traffic for one set of VLANs is sent in one direction and the traffic for the other non-overlapping set of VLANs is sent in the other direction. Therefore VLANs 1-100 are sent from SW1 counter-clockwise and VLANs 101-200 are sent clockwise. The purposefully blocked ports at points A and B only block half the VLANs each. If there is a real cable fault affecting point A, that would affect VLANs 101-200 which use this link. The REP failover would result in point B unblocking for VLANS 101-200. The switches SW2, SW3 and SW4 must flush their MAC address tables for VLANS 101-200 and relearn addresses on those VLANs. Meanwhile flooding needs to occur. A fast way to do this would be to extend the concept of an epoch per VLAN to an epoch per VLAN group. In this example there is one group for VLANS 1-100 and another group for VLANS 101-200. A way to increment the...