Each Performance Agent is assigned a manager Distribution Server when it is installed, or first connects to the Performance Domain. One of the responsibilities of the managing Distribution Server is for the Performance Data Engine (PDE) portion to assemble a superset of cube data for machines it manages.
To allow for high availability, the Distribution Server PDE provides redundancy. This is achieved by configuring each Distribution Server to replicate its Performance cubes to another Distribution Server, so that if the primary Distribution Server fails or becomes unavailable, the backup Distribution Server will have all cube data required to answer queries.
Once replication has been configured, the agent automatically delivers the cubes to both the primary and backup Distribution Servers. Additionally, the primary and backup Distribution Servers exchange cube lists, allowing the primary and backup Distribution Servers to pull cubes from each other. This ensures that both the primary and backup Distribution Servers contain a full superset of Performance cubes for machines managed by the primary Distribution Server.
When configuring Distribution Server replication, you must ensure that you use the correct server name in the cfgutil command. Failure to specify the full server name will cause the cfgutil command to fail and replication will not occur. The following procedure is considered best practice.
To configure Distribution Server replication
camstat -n
This command returns the correct server name to specify in the cfgutil command.
cfgutil -P <primary Distribution Server> -b <backup Distribution Server>
Replication is configured.
pdectl -h <backup Distribution Server> status
Status details for the backup Distribution Server are returned. The Backup for value should be the primary Distribution Server.
|
Copyright © 2010 CA.
All rights reserved.
|
|