Netbackup for VMWare

[ajalboush]

Hi,

I have netbackup 7.5.0.4 and I am backing up VMware environment, the problem is when I run many jobs together (3 or 4), I noticed that the size on the datastore (VM storage) become full, i checked and found that the snapshot take the same size as the primary copy, so whats wrong here, the size of the snapshot is supposed to be a few gigbytes, why it take the same size, it seems it do mirroring for the disk.

[Yogesh9881]

Need to know more details about your backup infrastructure ....... is it Vstorage API ??

[Yasuhisa_Ishika]

How did you check snapshot size? By DataStore brower or anything else?
What type of storage did you configured your DataStore on?
How about when you create snapshot manually in VMware?

[ajalboush]

Yes it is.

thanks

[ajalboush]

yes i check the size from datastore.

its EMC VMAX SATA drives.

I will try to create it manually and update you. 

[Mark_Solutions]

Are you using Block Level Incremental Backups (BLIB) and do you have change block tracking enabled - these will help reduce the size of your incremental backups

[ajalboush]

I enabled the BLIB, but my question is why the snapshot take size as the primary storage it should not exceed a few gigbytes? 

[Anonymous]

Can you clarify what you mean by Primary copy and Primary storage.?

Also, confirm there an no existing snapshots for these Virtual Machines. Look in the snapshot manager for each.

You are not confusing the size of the the original vmdk files with the provisioned size against the redo vmdk files (while a snapshot is active) - as seen in the Datastore browser and within the VM folder.?

[ajalboush]

I attached a sanpshot from the VMware mgt, the JAWESRS.vmdk size is 83 GB, and the JAWESES-000001.provisiond size is 83 GB, in VMware managment console i see the free space reduced by 83 GB when there is snapshot for this server.

there are no other snapshots. 

[Anonymous]

Exactly, the space taken up on the datastore at that point is your VM and subsidiary files + the 17MB file.

NOT the additional 83GB. This is thinly provisioned.

So it is not a problem.

[Yasuhisa_Ishika]

I agree with Stuart. It is normal.
When you create snapshot manually in VMware, you will get same result.
'Provisioned Size' represent logical size of vmdk, and 'Size' represent actual size of storage area aquired by this vmdk. Snapshot size is only 17MB.

[ajalboush]

Okay, but when I check the total free space on the vCenter I see that it is reduced by 83 GB and 17 MB.

what I Have is 8 TB for the datastore, I have free of them 3.56 GB free space, when I create the snapshot I see the free space become (3.56 - 0.83 = 2.73), so I see my free space become 2.73, and I have about 50 vitual machines, I am afraid if there are many snapshot at the same time that the free space become 0 then I face the virtual environment to hang. 

[Yasuhisa_Ishika]

i'm not sure what you see, but, assuming 3.56 "terabytes" datastore, i wonder why its free space reduces to 2.73TB(reduced by 830GB!). This topic is out of NetBackup , snd you should ask in VMware forum providing screenshots of vCenter and Datastore Browser.

[Anonymous]

Can you upload a screenshot of WHERE you see this free space result. Also detail version of ESX.

First off this is more a VMware problem, than NetBackup.

I would recommend also not to put all your VM's on one datastore. Splitting VM's and even VMDK's between datastores can help I/O queueing down the controllers.

Furthermore if you implement Storage DRS you can mitigate the issue of running out of space as VMDK's are moved around your datastore cluster as certain thresholds are met on datastores and therefore will not have the issue of running out of space and inevitably hanging guest VM OS.

EDIT: forgot to add you also have to add 4GB for your snapshot of the virtual memory. During snapshots the 'redo' VMDK' files grow at 16MB chunks based on the amount of change within the VM.

[Anonymous]

Have you resolved your issue, and/or share any insight as to what was going on?

And if any forum response helped or assisted you either vote or mark as solution.

Stuart.