SAN Question

[vmtech123]

Looking into a new SAN and I have a few questions.

First some info. 24/7 shop so there is no "outrage windows"I have many large DB/File servers ranging from 10-30TB. I run reverse incremental as I am doing restores frequently and want to most recent backup at my fingertips. I sometimes find it takes a substantial amount of time to re hydrate the backup. I am using Fiber and backing up from storage snapshots as we had some serious VM stun on the large VM's. That has improved with the storage snaps. Using a few V7000's with a few hundred TB each

My questions are:

How many IOPS or Throughput are required to run Surebackup or Veeam Labs? Surebackup takes days on these large VM's.

Is all flash a common config for a Veeam primary SAN these days? For those of you that are using Easy Tiering with say 40% flash, does it cause issues as I'd imagine most backup data isn't touched and will go cold. Only the changed blocks will be hot in the full backup if you have a lot of backups taking up space.

How has NAS backup been working for you? I plan to move the file servers to NAS backup and backup only the OS VMDK, unless there is an option to do both at a decent speed.

I plan for faster disk for the primary to do Surebackup / Labs on, larger slower disk offsite for archiving, and tape for long term.

Also, in regards to new servers, I know cores are where it's at, but with 20 cores/40 threads and a clock speed of 2.1 lets say, is clock speed going to be an issue?

And my last question, on average how many jobs do you run at the same time? I like to have more VM's in my jobs for better dedupe ( lets say 10-20 smaller VMs of the same os), but is it worth it to run more jobs in parallel to keep the windows shorter?

[foggy]

Hi Scott, in relation to slow SureBackup jobs, could you please share what is the backup target in your environment since this is what matters most for performance in this case - repository IOPS, not the primary storage.

[vmtech123]

So I just ran a 5 minutes quick test. I can try and push it harder later

Target is a v7000. I can hit at least 40k iops on I/O meter running several random tasks. 50% read 50% write.

[foggy]

Are you using SureBackup from storage snapshots?

[vmtech123]

Sorry, been very busy and my backups are working, but doing some digging on performance today, running something like a health check on a 12TB VM is at 80% after 12 hours. My SAN and proxy are practically idle.

Tried I/O meter again and it looks like my SAN has a ton of room, and the connection to the proxy seems to be good.

On my previous performance issues, yes, I am backing up from storage snaps.

I can get a few hundred K IOPS out of production and about 40-50k iops on my repo.

[foggy]

Are you talking about the backup health check? I still don't fully understand whether you're trying to do SureBackup right from the storage snapshots or from backups created from the storage snapshots (where backups physically reside on the backup repository, not the primary storage). If it is the second, then repository random I/O capabilities are most important.

[vmtech123]

I have mediocre performance doing both the backup health check and SureBackup on my repository. My backups are created from storage snaps on to our repo, then I use an immediate copy job to a second repo at another site. When I run I/O meter from either proxy (also the repos) the performance looks fantastic.

[foggy]

The I/O meter profile still might be different from the actual Veeam B&R workloads. You can contact our support to take a closer look at your environment, they might advise on possible optimizations.