Change veeam Repositorie on job ?

[tfloor]

Hello,

At the moment we are backup with veeam 6 p3 to a DD610.
We have bought a new QNAP highest model in range.
Now i want to change the Target in the job properties, but it said i have to move the backups first.

I don't want to do that, hold the old backups on the DD, and begin again on the new target (QNAP),

Is it possible to change existing jobs to a new target without moving the backups or create new jobs?
I don't want to move the files because the old target has very different job settings, (like no compression, because of dedup appliance).

BTW, i was thinking the QNAP must be mucht faster, but it's not. The bottleneck is not target anymore, but now it's source.
That's an EMC VNX SAN. The speed i get is between 80MB/S and 120MB/S. Is that normal?

Thanks.

Tristan.

[Gostev]

Is it possible to change existing jobs to a new target without moving the backups or create new jobs?
I don't want to move the files because the old target has very different job settings, (like no compression, because of dedup appliance).

No, in that case you have to create the new job.

BTW, i was thinking the QNAP must be mucht faster, but it's not. The bottleneck is not target anymore, but now it's source.
That's an EMC VNX SAN. The speed i get is between 80MB/S and 120MB/S. Is that normal?

Sure, these numbers sound like a single 1 Gb link fully utilized?

[tfloor]

Gostev,

I will recreate the jobs,

Regarding to the speed, i think it's not correct in my enviroinment.
I have a 4GB link (4 network adapters with link aggregation) between the physical DL380G5 veeam backup server and the Highest Qnap in range.

If i normally copy a file within windows to the qnap i get speed around 450, while veeam will not going further than 120.
I had a DD610 before, ( a normally slow appliance), but the speed was around the same as the high end new qnap...
Do you think there is something wrong?

Regards,

Tristan.

[tfloor]

BTW there is a switch between the backup server and qnap nas.
Anything i can test to get higher speeds?.

[Gostev]

Possibly, for example your link aggregation does not work, or your SAN cannot do reads faster than that, or something else like that. The backup target's performance does not really matter, because you said the source is being reported as a bottleneck. This means the job spends most of the time waiting for the requested data blocks to be read from the source storage. Thanks!

[rgj]

Hi,

"Workaround" I've used instead of recreating the jobs.

Select "Remove from backups", and then change repository. The next backup will be a full, to the new storage.

If you need the old backups, you can import them.

/René

[tfloor]

Possibly, for example your link aggregation does not work, or your SAN cannot do reads faster than that, or something else like that. The backup target's performance does not really matter, because you said the source is being reported as a bottleneck. This means the job spends most of the time waiting for the requested data blocks to be read from the source storage. Thanks!

OUR SAN is a brand new EMC VNX!
and the bottleneck before (with the dd610) was the target, that's why i decided to buy a qnap.
But i think there is something wrong with the link aggregation.

Can you maybe help me how i can test that?

[Heiko]

Hi

Just a basic explanation below as to my understanding
LACP is only utilized fully if you have multiple sources/target I/O requests at the same time.
Example in your case: An LACP config for the NFS share and connect multiple ESXi hosts ot is. Here you have multiple I/O request from the ESXi host going to an single target. More paths thus more I/O request can be handled at once byt the VNX. Not to forget of course to enable jumbo frames.
Meaning if you only have a single job it will only use one link at a time per I/O request. If you got two jobs then 1 I/O request can go via LACP pathA and 2nd I/O via LACP path2.
If your jobs are big then why not create multiple jobs over multiple Veeam Proxy servers (Virtual Machines).

Hope it make sense? Did increase my performance big time

How test? I would delete the LACP config or remove all links and leave only single link active and run the job.

[tfloor]

Heiko,

You have experience i see.
I'm working in SAN access mode.
With only 1 gigabit cable in my qnap i get around the same speed as 4.
And if i run multiple jobs at the same time with the same proxy / repository the speed is divided by the 2 jobs. (result the same 120MB/s)

The switch is a HP 2848, with a trunk with the command:
trunk 1-2 Trk1 LACP (the 4 ports to physical veeam server)
trunk 1-2 Trk2 LACP (the 4 porst to QNAP)

Please help me, let me know. (also if you think my speed must be more)
Thanks.

[tfloor]

I like to know what's a normal speed for my setup.
Maybe i want to start a new post, with a small setup description including their performance (MB/s, processing rate).

[Heiko]

As Gostev mentiond the speed you get per 1GB link is very good. Just do a google to see what performance you only can get on a 1GB link .
The job is halfed because of the singel source I/O request. Try to 2x VM Veeam Proxy server for testing. Change the two jobs to use different proxys and test in Virutal Mode.

[tfloor]

Heiko,

You are talking about 1 GB link, but i'm using a trunk of 4x 1GB link.

I've also tried to do a seperate job with a VM as a proxy (windows 2008r2), then it's using hotadd method, and that's get me speed around 25MB/s.
So that's not an option for me, very slow.

So again, i'm using a 4GB ethernet trunk.
That's must make sense in speed!

[tfloor]

Is it possible that the speed problem is because of a incorrect ESC vswitch / physical nics teaming method?
Before my target was the bottleneck and now only the source and proxy.