Polycom RealPresence Resource Manager System Operations Guide System Redundancy Overview
518 Polycom, Inc.
How Redundancy Works
Terminology is very important in understanding how redundancy works.
In a redundant configuration, one server is licensed as the primary server and
the other server is licensed as the redundant server. The primary server is
always the primary server and the redundant server is always the redundant
server.
In a redundant configuration, there is only one active server and only one
inactive server. The active server is the server currently managing endpoints
and monitoring conferences. It is actively responding to network traffic routed
to the virtual IP address. In a normal operational state, the active server is the
primary server. In a failover state, the active server is the redundant server.
The active/inactive servers over a dedicated private network using a LAN
cable physically connected from one server to the other. They communicate
every 200 milliseconds using a private IP address and port 5405. If the inactive
server does not receive a heartbeat from the active server, it will promote itself
to being the active server.
The most common reasons for system failovers are power failures and
network disconnections. Failures in services running on the primary server
also initiate a failover.
If both the primary and redundant servers start simultaneously (for example
if both are in the same location and recover from a power failure at the same
time), both servers will initially attempt to become the active server.
Whichever server starts first becomes the active server.
An administrator can force a failover via the Switch Server Role function in
the Resource Manager system user interface.
Also, the failover to the redundant server occurs seamlessly because the
endpoints are registered with the virtual IP address, which remains constant.
However, endpoints that are dynamically managed will lose the connection as
the provisioning service will stop for approximately five minutes.
During a failover:
• Users logged into the RealPresence Resource Manager system user
interface are disconnected during a failover and returned to the main
RealPresence Resource Manager system web page. Users can log back in
once the failover is completed.
• Users in the middle of an operation may get an error message, because the
system is not available to respond to a request.
• The redundant server becomes the active server. Its services start in an
order designed to prevent the new active server from being flooded with
requests from endpoints during startup.
A system failover usually takes approximately 5 minutes.
