04-18-2013 08:22 PM
We're currently deploying a new version of TRIM (7.31) within our organisation. As we have a pair of primary sites with replicated storage (SAN), dark fibre etc. we intend to use multiple workgroup servers for load balancing and mirrored SQL 2008 R2 databases for additional disaster recovery functionality (on top of the usual backups).
In the spirit of "keeping it simple" the intention is to setup the mirrored database in manual failover mode with no witness. So if our primary data center goes off the air, we merely need to stop TRIM services, point the still available workgroup servers connection string to the mirror database instead, make the mirror database the principal one via the usual MSSQL procedures and restart TRIM services. This model did appear to work in my small test environment.
From a TRIM perspective, it should always be functioning in its supported state (one RDBMS and multiple workgroup servers) even though the RDBMS is quietly replicating data in the background to it's mirror without any HP TRIM knowledge or interaction.
Can anyone see any glaring problems with this setup? I deliberately stayed away from automatic fail over and/or tampering with the correctly functioning connection strings as "high availability" is not a requirement and I want absolutely no possibility of HP stating that this is an unsupported configuration.
This post is also admittedly light in detail but I'm just looking at the overall scenario as well as thoughts from users who may have explored or implemented this type of build in the past.
Thanks for your time.
05-05-2013 08:09 PM
We have been running for nearly 5 years with TRIM 6.25 on a SQL Mirror with a witness (high availablility required) and it has generally been fine. Have had the occassional undesirable failover but nothing we couldn't sort out. Upgrading to 7.2 as we speak and will keep the same SQL mirror solution