We have a new 6.5 installation. Veeam server backing up vmware to a HP D2D backup device over the network.
The D2D does deduplication so we set veeam backups to not do any dedup or compression.
We've set the vmware proxy but backup performance to the D2D is very slow.
I am new to veeam so I'm hoping you all have some good feedback on areas I can look at to boost performance, whilst maintaining good dedup ratios on the D2D.
Don't disable Veeam dedupe, it does not affect more advanced storage dedupe yet improves backup performance noticeably.
Other than that, deduplicating storage devices are primarily intended for archiving and thus make it pretty slow primary backup targets, so don't expect your backup to fly when writing directly to this sort of storage... I usually recommend backing up to fast raw disk and only keeping a handful of restore points there, but then copying backup to archiving storage such as HP D2D and using that for long-term retention.
chrisdearden wrote:Its faster to do a weekly active full usually.
Depends on change rate and full backup size really, so definitely not a universal recommendations... I've heard about environments where active full would take a few days (weekend was not enough), but synthetic would complete within 1 day (change rate was minimal, so incremental backups were pretty small - so synthetic full creation was 90% sequential I/O, good for dedupe devices).
There are no snapshot limits, as they are per-VM not per-datastore with VMware backup. As for the best practice regarding the number of streams hitting the target storage, this is storage-specific, so you should check with your backup storage vendor directly for recommendations. Thanks!
Mike, you can create multiple jobs that will write to the same target simultaneously. There is also a setting allowing to control data ingestion rate available in the backup repository settings, which limits the write speed to the specified value.
the solution thats been designed is that we backup the complete datastore to the D2D CIFS fileshare.
Are we right in saying that the only way to increase throughput would be to add more proxies?
At present we have "vmware backup proxy" but if we added a physical server proxy too, will the one backup spread the load across multiple proxies?
Another question....
If you run the same backup within 24hrs it tries to do an incremental. How can we override it so that it does a full each time?
Unfortunately this customer wants fulls every day....
maverick964_uk wrote:At present we have "vmware backup proxy" but if we added a physical server proxy too, will the one backup spread the load across multiple proxies?
Since VMs in a job are processed sequentially, you still have to create multiple jobs to spread the load between proxies.
maverick964_uk wrote:If you run the same backup within 24hrs it tries to do an incremental. How can we override it so that it does a full each time?
Unfortunately this customer wants fulls every day....
These are two different scenarios, actually. If he needs active full backup every day (so that full source VMs data being retrieved from the storage each time), you can schedule it in the advanced job settings. If he needs full each time the job is run (even several times a day), then this can be achieved via PowerShell, I believe.
