Backup Copy with AltaVault

[jeff.richeson]

We use the NetApp AltaVault as a secondary target for backup copy jobs. The backup copy job runs nightly with monthly fulls for long-term retention. As recommended, we have selected "read entire restore point for source backup". It appears that some amount of data is required from the latest monthly full. This becomes a problem when the latest full is only in the cloud and no longer in the local disk cache. The backup copy job is delayed as the AltaVault has to populate the local cache. What data is needed for the latest full and any thoughts on how to avoid pulling data from the cloud?

[Dima P.]

Hello Jeff,

It appears that some amount of data is required from the latest monthly full.

Any chance you've enabled GFS retention on this backup copy job? Thanks!

[jeff.richeson]

Yes, GFS is on.

[Gostev]

AFAIK there's no way you can make NetApp AltaVault work well with Veeam.

[jeff.richeson]

Thank you for the honest answer. That is the conclusion I am coming to. While things mostly work, they don't work well with the combination. Unfortunately, both companies publish whitepapers on the proper way to get Veeam and AltaVault configured to work together. I know other people are attempting to use Veeam with the AltaVault. This is the hardware and software I have so I have to make the best of it. Are there any settings in Veeam to minimize the reading that is done from the target repository when doing a backup copy?

[foggy]

Basically Veeam B&R shouldn't read anything from target in case "Read from source" is enabled. There's a chance it is reading metadata from cache, but not monthly full itself. To verify that we need to investigate logs. Btw, what type of backup chains do you use - single or per-VM?

[collinp]

We backup using veeam directly to the altavault with daily incrementals and monthly active fills. Didn’t have very good luck using backup copy jobs or synthetic fulls though since this is a dedup appliance and these caused a lot of I/O on the AV. You may need to purchase more local disks for the altavault so the latest backup chain remains in cache and you aren’t forced to pull files from the cloud.

[jeff.richeson]

Foggy, I apologize for the delay in responding. Apparently, I'm got receiving notifications of replies. I'll look into the settings more.

I have a support case: 03344531 but we haven't made lots of progress. It isn't for a lack of trying from the support engineers. My license doesn't provide per-VM so I'm using 15 backup copy jobs to keep the backup file sizes reasonable. When I asked the NetApp engineers to look at what files where being requested, the gave names of the GFS fulls. Looking at the amount of data read (~20GB daily) it is obvious that it isn't reading the entire restore point. Multiple reads seem to really bog down the AltaVault causing the NFS to become non-responsive for some time.

[foggy]

My license doesn't provide per-VM so I'm using 15 backup copy jobs to keep the backup file sizes reasonable.

Then this looks like the cache is being read.

[jeff.richeson]

When you say the cache is being read, you mean the files in the repository (the AltaVault)? Does per-VM backups eliminate the need for the reads?

[foggy]

Yes. Yes, the cache mechanism is implemented for the case where per-VM chains are not used.