Polycom 3725-76302-001LI The Polycom DMA Systems Three Configurations

Introduction to the Polycom DMA System Polycom® DMA™7000 System Overview

Polycom, Inc. 5

The Polycom DMA System’s Three Configurations

The Polycom DMA system is designed to be deployed as a pair of co-located

redundant servers that share the same virtual IP address(es). The two-server

cluster configuration of the Polycom DMA system has no single point of

failure within the system that could cause the service to become unavailable.

The two servers communicate over the private network connecting them. To

determine which one should host the public virtual IP address, each server

uses three criteria:

•Ability to ping its own public physical address

•Ability to ping the other server’s public physical address

•Ability to ping the default gateway

In the event of a tie, the server already hosting the public virtual address wins.

Failover to the backup server takes about five seconds in the event of a graceful

shutdown and about twenty seconds in the event of a power loss or other

failure. In the event of a single server (node) failure, two things happen:

•All calls that are being routed through the failed server are terminated

(including SIP calls, VMR calls, and routed mode H.323 calls). These users

simply need to redial the same number, and they’re placed back into

conference or reconnected to the point-to-point call they were in. The

standby server takes over the virtual signaling address, so existing

registrations and new calls are unaffected.

•Direct mode H.323 point-to-point calls are not dropped, but the

bandwidth management system loses track of them. This could result in

overuse of the available network bandwidth.

•If the failed server is the active web host for the system management

interface, the active user interface sessions end, the web host address

automatically migrates to the remaining server, and it becomes the active

web host. Administrative users can then log back into the system at the

same URL. The system can always be administered via the same address,

regardless of which server is the web host.

The internal databases within each Polycom DMA system server are fully

replicated to the other server in the cluster. If a catastrophic failure of one of

the database engines occurs, the system automatically switches itself over to

use the database on the other server.