Help on Veeam config with fiber channel backup storage?

[davidb1234]

I am designing a new vmware virtual setup for our public facing web environment. Since Veeam is working so well in our corporate environment and for DR I would like to also use it there.

My plan is the following:
-1 Dual Channel Fiber Channel SAN Enclosure running 15k fast disks in RAID 10 running all productions VMs.
-2 Fiber Channel Switches setup with multipathing
-Another Dual fiber channel Channel SAN Enclosure running 15k fast disks in RAID 10 storing veeam backups. This would be the Veeam repository.
-1 physical 12 core Veeam Backup Server attach with fiber channel to both SANs. The backup SAN would be partitioned with NTFS for Veeam repository.

So basically I will be backing up directly from production SAN to backup SAN over 10GB fiber channel through the physical backup server.

Would this give me insanely fast backup and restore speeds? The other idea would be that in an emergency we can use the 2nd backup san to do instant recovery if our production SAN failed(both controllers failed or something worse like a backplane) until we can get it repaired and then vmotion them back over to the production san from the backup san.

Is this just overkill or is there any bottleneck with Veeam that I am missing?

[Gostev]

This spec should be flying. But yes, your backup SAN is probably an overkill for hosting backups... lower RPM disks should be fine, remember Veeam compresses and dedupes data at source reducing it at least by half. For that reason, I would change small high RPM disks to bigger low RPM disks, as disk space is more important for backup repository than speed (will allow you to have longer retention policy).

[davidb1234]

In my current production and DR environment our TARGET storage is always the bottleneck on all of our jobs. Is that not usually the case? They are all RAID 6, direct attached storage to the physical backup server. I figured trying SAN attached RAID 10 would help with backup speeds and allow us to run the VMs in an instant recovery mode for a day or two if necessary.

Long retention is not as important as fast backups and quick recovery and possibly running in instant recovery mode with multiple vms at the same time for a day or two.

[Gostev]

Are you using reversed incremental backup?

[davidb1234]

Are you using reversed incremental backup?

Would not ever go any other way. We love having full VM backup files to restore instantly all the time. Synthetic fulls are always turned on. I have just never tried using Fiber Channel attached storage for the backup storage and fast disks. I have always gone as you suggested with cheaper larger disks and direct attached storage.

[Gostev]

Well, the restore performance does not really depend on the backup mode. In fact, you will find that in the long run it will be slower for reversed incremental backup mode due to VBK fragmentation, unless you are doing periodic active fulls.

So the choice is, basically, between these two options:
1. High RPM drives are best for I/O intensive reversed incremental. They are smaller, but reversed incremental is disk space efficient, so good match.
2. Big low RPM drives are best for forward incremental. They are bigger, so good match considering storage consumption due to periodic fulls requirement.

Unless you keep copies of your backup, forward incremental backup mode is actually preferred from reliability perspective (since you always have multiple full backups available). And, forward incremental VBK files also do not suffer from fragmentation problems, because the are created from scratch weekly.