CMS Architecture - High Availability

HP CMS support application clustering. Refer CMS sizing guide for performance numbers from RnD lab machines to understand how CMS solution perform in various configruations.

Facts on HA - Application clustering model

• External Access to UCMDB Servers via the load balancer, where two or more UCMDB Servers are configured.

• All of the UCMDB Servers are active at any given time & can handle both read and write requests.

• Requests are distributed to the UCMDB Servers in the cluster by the load balancer.

• All read requests are shared evenly among all of the UCMDB Servers (Readers), only one UCMDB Server (Writer) is also responsible for write requests at one time.

• Any write requests received by a Reader are passed to the Writer.

• Any of the UCMDB Servers can take over the Writer role in the case that the Writer becomes unavailable.

Load balancer Requirements

• Cookie Insertion (in all responses to clients)

• Health Monitors

• (For SSL) Redirect Rewrites

HA Logs

• cmdb.ha.log – short log. Displays info messages in short format. In case of exception only the message appears.

• cmdb.ha.detailed.log – also has exception stack trace. Has much more details in TRACE mode. For example: each HA request has request ID, and we can follow the request’s trace between the servers.