Hi
I have a situation where we are backing up using accelerator and using change journal in a clustered environment. In testing I've enabled "Use change journal" on each node of the cluster and run backups from a client virtual hostname from the second node. Full backup no change journal used as expected, first backup will take a little time. Subsequent backups fine, using change journal, finishing in seconds, both inc and full, due to very little change on the host. The track log has been moved to a cluster aware drive that is accessible from both sides, which means there should be no issue with using it, and certainly there are no errors reported.
The problem comes when we failover. The change journal is now no longer valid as it's a completely different host, so we have to go through the same "initialisation" process, for want of a better term, every time resource fails over - an accelerator backup with no change log, then subsequent backups will use it.
This is all very well for a small environment with a few files but a) we wouldnt necessarily use accelerator with smaller volumes and b) accelerator is targeted specifically at high volume, high change environments, so it would become intergral to backup operations
The questions are therefore
1. Can accelerator be made in any way aware of clustered environments, as far as the change journal is concerned?
2. Using the change journal with accelerator makes such a massive immediate difference, at least as it appears to me in testing, so would using accelerator over a peroid of weeks and months eventually return the sort of numbers we see with change journal enabled?
3. Slightly off topic but related, we have multiple "clients", i.e. cluster resources on separate IP addresses, sitting across our cluster, living in several policies. Best practise says not to have multiple policies due to conflicts with the USN, but it would be inconvenient to have just one policy for all of the cluster resources, not least from a reporting point of view. Does this conflict arise when backing up the same thing, or different resources of the cluster?
Currently I can't see a reason to use accelerator at all unless change journal can be made consistent across a cluster, or accelerator steps up to the mark without it!
Anyone with any experience or advice in this area ? Any info would be greatly appreciated!
Thanks in advance
Chris