05-28-2012 03:16 AM
Hi,
I have some backup jobs running over a weekend, when there is no activity on the network, both jobs backup from the same remote server, one backing up drive x and one backing up drive y. The drives are both iscsi targets to the same san, so essentially identical jobs (with differing data).
One starts before the other and runs at at 2400mb/minute, the other at 1600. I can see that this could be the case as the first is taking the bandwidth. However when the first job finishes, the other does not get any faster - which seems a little strange to me given they are using the same hardware/software/network path and the bandwidth has been freed up with the first job finishing.
I see similar results with both jobs starting at the same - more consisten sizes, but the second one not getting any faster?
Is there any reason for this?
thanks in advance for any thoughts
05-28-2012 03:37 AM
Hi,
The data rate will not necessarily increase in speed. It's dependant on what you're backing up in terms of size, data type, network throughput of the targets, your iSCSI target (stripe size on the disk, RAID level) etc.
Check these and see if there is any change. The other thing to test is to schedule the slower job to run 10-15 minutes earlier than the job that runs faster...
Thanks!
05-28-2012 05:49 AM
Hi, thanks for your reply, it just seems logical to me that when the resources and bandwidth are freed up by one job stopping, the other job should make use of them, especially when they're sharing all those resources and bandwidth in the first place.
The targets have the same settings, same raid, same size etc and the same throughput when copying files. The backup job is running from the same server to the same server over the same network and interfaces.
I wouldn't necessarily expect the seond job to suddenly jump to 4000mb/s but would expect to see some increase??
05-28-2012 06:46 AM
It would seem that way...but try that test and swop the jobs around and see if the results stay the same or not...
05-28-2012 06:31 PM
My experiance (BE 2010 R3) is that a single backup job will "never" use all the bandwidth, the BE services cannot request or process the data fast enough.
A number of backup jobs will utilise the full bandwidth.
I have come to this conclusion by copying a 500Gb file from a remote server to the BE server in the same share that the backup to disk storage is.
That file copy can soak 80-90% of the bandwidth
Backing up that same file using backup exec and a BE Agent on the remote server utilises about 30% of the bandwidth