Has anyone come across a tremedously slow vmware backup when a VM has 2+TB vmdks attached? I am having this issue with two VMs. The backups are going through the SAN in a NetBackup 7230 3.0 version. Another info, sometimes the backups end with status 50 (client process aborted)
Any comments is very much appreciated.
slow backup could be because of many reasons, and its hard to find it for vmware backups.
if it is possible, first thing i would suggest you to check backup speed and time with NBD transport for testing, it would help to narrow down if there is any perfromance issues in SAN side.
I am also interested to know the VMware enviornment versions., does it enabled with CBT?
I see your point. What really catches my eyes is that this is happening with just this two VMs. And they all have some vmdks with the size of 2 TB or more. But I will see if I can test it using NBD transport mode. I believe I have already tried that but I will double check it.
when you say same VMDK, are they really same VMDK shared with Two VMs? or same type of VMDKs with independent to each VM?
Same type of VMDKs with independent to each VM. See attached the report of one of the backup job that is taking forever to finish.
Just to add, this behavior is also happening with a VM that is the file server. And in this case this VM has a size of 16TB. Each VM sits on a separate datastore and all of the LUNs are share with a NetBackup 5230 appliance with two 8Gb HBA ports, with each port connected to a separate fabric switch.
16TB, i would not even try to backup this size VM though VMware snaphost backups method, in case if we need a backup of VM recovery i would go with boot disk backup only ( or specific VMDK backup which is avaliable now in 8.1).
lets focus on those 2 VM.
i would like to see the results which will help to narrow down the issue
1) NBD backups speep
2) SAN backup speed without selecting "Enable filerecovery from vmware backups" & "Enable block level incr backup" from the Vmware tab of the policy.( this is to make sure that file cataloging is not kiling the speed)
3)some numbers, when you say slow backup speed, what is the Kb/sec you are seeing for these VMs and what is the speed on other VMs.
For the question no 3, the speed is roughly compared to a fast ethernet, around 10MB/s. The backup for this particular server is already going through the netbackup client. However, there is another machine with 2TB disk size that is also presenting the same issue. The backup stays running for hours long and suddenly ends with status 50. I have also configured the policy to use the netbackup client to send the data over to the media server. This method work smoothly but if I have to recover this machine I will have to use baremetal and this feature might take a little longer than a snapshot restore. What are you thoughts on this?