Backup throughput

[dstevenson]

Greetings,

I am seeking confirmation / information of a sort, I would like to get the best
throughput I can for my nightly backups.

Storage: Primary is XIV with gobs of spindles
Destination: V5000 with about 70 TB space
ESXi 5.5
Veeam 7 update 4
Fiber attached storage
Proxy configured as Virtual Appliance
Parallel processing is enabled
Each proxy is configured with the following specs
8 cpu
12 GB ram
Win 2008 R2
Transport mode is Virtual Appliance
With failover to network enabled to ensure backup is completed
Connected Datastore is Auto
Max 8 concurrent tasks

Watching the proxy performance I am not showing any kind of taxing on the cpu or memory

When I review my job they each say Target for bottleneck but I ask the storage
team and they show me reports that the storage isn't doing much. Any ideas on
how to increase my job through put.

Has anyone had any trouble with AV slowing things down
Size of LUNs
Any general thoughts would be helpful,

Thank you

Dana

[foggy]

Dana, how your target storage is added to Veeam B&R console (what type of repository it is) and what backup method do you use in your jobs (forward or reverse incremental)?

[dstevenson]

repository is local windows on a FC connected storage IBM V5000

Reverse incremental

[foggy]

You can try to switch to forward incremental, since it puts less load on the backup storage. Btw, what performance do you currently get? Note, that there always be the bottleneck, with every setup.

[dstevenson]

I am trying to change to forward and see if that helps.

Throughput is about 20MB/ps but can burst over 200 MB/ps. I am monitoring the storage, no performance issues yet from what i can see.

[dellock6]

The bottleneck can be at the raid controller in some cases, or also at the spindles speed. Since Reversed is heavily random, disks have to move back and forth on their plates, and they have latencies in doing so. Unless you have really low-level monitoring, it's hard to see this in storage monitoring tools. but I'd say 99.9% of the times, on REversed bottleneck is always on the target due to the nature of this mode.
You will for sure see huge improvements with Forward Incremental.

[dstevenson]

So I did a forward job and BAM 1140 MB/ps avg. Makes for a happy VM admin. I changed the selection to Direct SAN Attached also and disabled the network fail over. Thank you for the help and information ....

So the next question would be what job do I run to tape, file for the repository folder or job. If I do a job to tape will it backup the entire set of files needed to restore every night or do i need to keep multiple tapes?

I also looked at the file level backup so just to backup that folder that the backups reside for that job.

Again thank you so much for the information...

[veremin]

If I do a job to tape will it backup the entire set of files needed to restore every night or do i need to keep multiple tapes?

What type of tape job you're going to use? Files to tape or backup to tape one? Both are incremental, meaning that they will copy files or backups that haven't been copied to tape mediums previously. Corresponding section of our User Guide should clarify most of your questions. Thanks.