Director Container Storage Pools and Disk Space Used

[rowl]

TSM 8.1.0.0 on Linux.

I have some puzzling numbers that I am trying to understand. I ran a generate dedupstats for everything on my server, then when I select * from dedupstats I summed up the columes and see values like Protected = 200TB, Physical=40TB, Total Saved 160TB. So it appears around 80% savings and I am happy with that.

When I look at the TOTALSUROCC from the Status table it shows 200TB, and the used space in the file systems that make up my directory container storagepool add up to 200TB. Since this TOTALSUROCC is what we are licensed against I am concerned.

I was expecting the disk usage to be proportional to the deduplicated/compresses size of the data. So in this case 40TB (or maybe 80TB to account for 1/2 full containers) not the full size of the protected data.

The other puzzling thing I see here is that on the target server that I run protect stgpool and repl node to, the "Additional space for protected data" from q stg f=d is 5.5TB. Everything from the above source server is covered by the protect stg and node replication processes, so I don't see how I could only be storing 5.5TB there.

Does anyone have some insight into these values? Maybe we have a fundemental misunderstanding of the these, but I have not been able to find good definitions anywhere.

Thanks
-Rowl

 

[marclant]

I was expecting the disk usage to be proportional to the deduplicated/compresses size of the data.

There is empty space in containers.

The other puzzling thing I see here is that on the target server that I run protect stgpool and repl node to, the "Additional space for protected data" from q stg f=d is 5.5TB. Everything from the above source server is covered by the protect stg and node replication processes, so I don't see how I could only be storing 5.5TB there.

Does anyone have some insight into these values? Maybe we have a fundemental misunderstanding of the these, but I have not been able to find good definitions anywhere.

It's the latter, a fundamental misunderstanding

Additional space for protected data
The amount of space, in MB, that is used to protect data from remote servers. This is the total amount of space used for data received from other servers as a result of running thePROTECT STGPOOL command.

After the PROTECT STGPOOL command is run, the data is not assigned to a node. However, if you run node replication on some or all nodes, then the data is assigned to the nodes and is no longer assigned to the additional space for protected data.

If you do not run node replication, then the data received (after the PROTECT STGPOOL command is run) remains assigned to the additional space for protected data.
source: https://www.ibm.com/support/knowledgecenter/SSEQVQ_8.1.5/srv.reference/r_cmd_stgpool_query.html

Look at the Estimated Capacity and Pct. Util.

TSM 8.1.0.0 on Linux.

Should consider 8.1.5 or if you are not ready to commit to SSL, 8.1.1.100. Lots of fixes and enhancements that you are missing out.

 

[rowl]

Thanks for that tidbit, I don't recall reading that additional information when I looked up this....

However, if you run node replication on some or all nodes, then the data is assigned to the nodes and is no longer assigned to the additional space for protected data.

So with that info, this 5.5TB of additional space is what, orphaned data? Node replication runs daily for all nodes.

Planning to go to 8.1.5.0 ASAP, just the amount of red tape imposed in my environment makes doing anything a long slow painful process.

I know there is empty space in containers, and I have seen automated container move processes running from time to time. In the back of my mind this seems like tape or file devices where we would swag at 2x capacity to account for the empty space. 5x seems a bit much.

I have a PMR open now, see what they come back with.

 

[marclant]

So with that info, this 5.5TB of additional space is what, orphaned data?

Possibly.