WAN Acceleration Cache - Redundancy Needed?

[wa15]

Good morning all,

Is not having a redundant setup for the WAN accelerator cache an issue? If I just have a single 10K RPM disk serving as the cache and that disk fails, will that cause any issues besides having to pop in a new drive and letting the cache rebuild?

Also, a somewhat separate questions, is having the Guest File System Catalog and WAN accelerator cache on the same drive (but different folders) an issue?

Will appreciate some help!

Thanks all.

[Gostev]

Hi Farshid, you are correct: in case of drive failure, simply pop in a new drive and let the cache rebuild. The only issue I can think of is potentially missing your offsite RPO for the copy period following drive failure.

Having the Guest File System Catalog and WAN accelerator cache on the same drive should not be an issue as long as your Backup Copy jobs start once the primary backup jobs finish. This way, these two workloads will not overlap.

Thanks!

[wa15]

Awesome, thank you Gostev. Appreciate the quick reply

And I am assuming the same thing would be true for the vPower NFS path, correct? If the path is on a drive that fails, we can just replace the drive and we should be good?

[Gostev]

Yes, unless it fails while you have instantly recovered VM running off backup. In that case, you will lose all the changes happened since the VM was started, and this can mean significant data loss depending on how long the VM was running.

Additionally, you don't want to share vPower NFS drive with any other workload, because it is going to get hit hard by instantly recovered VMs.

[wa15]

Makes sense. And sorry about asking so many questions, but would having the vPower NFS data store shared with the location of the Guest File Index Catalog be okay? We have two 300GB 10K RPM drives in RAID1 dedicated to the Index Catalog. I know the performance would probably be on the slower side for vPower NFS, but any other issues?

This is a new deployment of Veeam Enterprise Plus (we are new customers so I am trying to see how to best do this.

[Gostev]

To my knowledge, catalog does not carry heavy I/O workload, so this might be okay.

[wa15]

Got it, thank you!