how to reduce used capacity in data protection?

[Rodolfo_Giovann]

A customer of ours uses vSphere Data Protection.

In the reports, I see that Used Capacity is growing at 1%~2% per day.

It has reached 86.31%.

We removed old backup some days ago, so it should have been able to complete the maintenance, but it is still growing.

How can we lower the used capacity?

Thank You very much and regards

Rodolfo Giovanninetti

[LennieDH]

I believe this is a regular behaviour of data recovery.

According to KB 1027606 space in backup destination is allocated using slab technique which allocates chunks of contiguos blocks.

When old backups are deleted the corresponding data blocks of memory are deleted but slab containing them are not unallocated. Empty areas inside each slab are marked as "free" for subsequent uses.

Hope this could help!!

Greets!!

[Rodolfo_Giovann]

Thank You for the help.

The article that You mention is related to data recovery, that I believe is the previous version, whereas the customer has data protection.

Anyway the basic concepts might still apply, and in fact I thought that maybe when I remove a backup the space is not completely and fully released, but is marked as free and is made available to be reused for following backup.

BUT, if it is the case, I should NOT see the used capacity that increases everyday.

For example, imagine that I reached 80%.

I remove old backup for say 10%.

So, now there is 70% used, 20% completely free, and 10% that will be reused by data protection again.

I should see that the used capacity stays at 80%, because data protection REuses freed space.

Instead, the used capacity just one day later my removing old backup goes up to 81%, then to 83%, and so on.

Are there commands that let me check these data?

Thank You again and regards

Rodolfo Giovanninetti

[LennieDH]

I assume that in your case the space occupation continues to grow due to the fact that VMware needs to allocate new slab files to store blocks. This is a typical behaviour of data recovery which tends to allocate all empty space in backup datastore until all the space is claimed (but not necessarly filled).

Reclaim operations are performed to move used memory blocks to continuous locations preventing gaps between blocks which implies that all used memory blocks are moved leaving all free memory block contiguous. This process, I assume, is performed without deallocating slabs so used capacity is continuously marked as growing by data recovery even if actual free space is increased.

KB 1039075 could be useful

...But, I repeat, these are just my assumptions I might be wrong on this...Anyway if this works as in data recovery even if space is marked as 100%  used you should be able to keep performing backups since older backup copies are automatically deleted and replaced by new ones.

Regards.

[markko]

DP Administration Guide says in page 48:

vSphere Data Protection capacity thresholds
Threshold               Value   Behavior   
Capacity warning        80%     vSphere Data Protection issues a warning event.
Capacity warning        95%     Tasks are not generated on vCenter for backup jobs when capacity is greater than 95% full.
Healthcheck limit       95%     Existing backups are allowed to complete, but new backup activities are suspended. vSphere Data Protection issues warning events.
Server read-only limit  100%    vSphere Data Protection transitions to read-only mode and no new data is allowed.

We hit the 95% and although I deleted older backups the space stays over 95% and Backup scheduler constantly refuses to start, that is because appliance got "health check limit reached event". No scheduler , no backups. Even if I acknowledge events the capacity is still over 95% and server gets again health check events and backup scheduler stops and refuses to start because of these events. I guess I must reinstall appliance or add another and start new backups because current appliance isn't able to perform backups anymore.

[jhunter1]

Please open a support request (SR) with VMware Support on these types of issues as it helps with a couple of things: 1) VMware Support can help you resolve the issue and 2) VMware can identify possible issues with code and get them fixed. Thank you.

[markko]

  Solved my problem with additional VDP appliance. Because problem was too much data for one VDP. No problem with reclaiming space(I kind of feel a little shamed because I didn't look at the total size of VM's which were backed up) but starting Scheduler service when space usage is over 95% is still problematic. At least I didn't find a way from GUI to get it work. Only from cli and with "mccli event ack" and after that Scheduler service started up again.

[jhunter1]

Thank you for the update. Good to hear you solved the issue. This may be of interest, as well:

http://blogs.vmware.com/vsphere/2013/04/vdp-5-1-10-ga.html