Browse Prior Art Database

A method of mirroring multicast state in hosts in wireless systems By George Popovich and Vidya Narayanan

IP.com Disclosure Number: IPCOM000005498D
Original Publication Date: 2001-Oct-09
Included in the Prior Art Database: 2001-Oct-09
Document File: 4 page(s) / 31K

Publishing Venue

Motorola

Related People

George Popovich: AUTHOR [+2]

Abstract

A method of mirroring multicast state in hosts in wireless systems By George Popovich and Vidya Narayanan

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 50% of the total text.

A method of mirroring multicast state in hosts in wireless systems – By George Popovich and Vidya Narayanan

Introduction

Future CGISS packet-based systems face the challenge of being highly available. For sake of high availability, it becomes essential to provide redundant devices in the system. Apart from hardware availability, protocol availability also becomes essential so that the backup device can take over the services of the primary device upon failure of the primary device. It is necessary to make this switchover as seamless as possible to the end user.

Systems that have a lot of multicast traffic  may need to preserve multicast state for sake of availability. In such a case, there needs to be a method by which a backup host can take over all the multicast services handled by the primary host prior to its failure. Similarly, there must be a method to transfer all the state back to the primary upon its recovery, if so desired. In the absence of such a feature, all the multicast flows would be lost when the primary device fails.

Background

In the case of routers running a multicast routing protocol, it is possible to maintain state on multiple routers on a network by simply listening to the IGMP Membership Reports and Leaves that are sent by hosts on that network. However, there is no method of maintaining multicast state in multiple hosts on a network. This is especially true if the same hosts originate the IGMP Reports and Leaves.

Currently, there is no known method to mirror the IGMP state between hosts for seamless multicast convergence on failure of the active host.  

Solution

All multicast state changes would be exchanged between hosts via a message triggered by the multicast state change, so that the backup host has all the state ready when the primary host fails. Upon recovery of the primary host, all the state will  be conveyed through a message from the backup host so that the primary can take over.

Consider two hosts (A and B) on a given network (subnet) N as shown in the figure below. Let us assume that Host A is the primary device and Host B is the backup device. A hello protocol would be run between the two hosts on the shared network. The frequency at which the hellos are sent would depend on the desired convergence time.

Under normal conditions, whenever Host A joins a new multicast group, it sends a “Multicast Status Message (MSM)” indicating the addition of the new group. Host B, on receiving this message, creates an entry for the group in its “Multicast State Table (MST)”. Similarly, when Host A leaves an active group, it sends an MSM to Host B, indicating the group left. This triggers Host B to remove that particular group from its MST. This way, at any given instant, Host B has a list of the all the active multicast groups. It is to be noted that although Host B maintains this list of active groups, it does nothing with that information as lo...