Unusual dedup information required

[DanGiles]

I'm running 7.1.4 on RHEL 6.6. Using FILE device class deduped storage pool and will be converting to directory containers in the not too distant future.

I keep getting asked how much data will be added to TSM for new servers. This is not at all an easy answer in a deduplicated environment. I think there may be a way to get this information, but not from any tables accessible to TSM. So I figure there should be DB2 tables where I can figure out the following:

Node A uses extents a,b,c
Node B uses extents b,d,e

So Node A has distinct extents a and c, but then I'd need to know the size of the extents.

Maybe there's another way of getting the amount of data reduction for a node? Then I could average the amount of data reduction of a given platform. Maybe the only thing I can do is scan the activity log for ANU2526I, but that really wouldn't tell me how much a NEW client is going to add to the storage pool.

Anyone looked at this? Am I spitting into the wind?

 

[DanGiles]

Or do I simply look at logical_mb vs reporting_mb? It LOOKS like that should give me more or less what I want (?)

 

[DanGiles]

and as soon as I post, I start finding things! If anyone is interested in stuff like this, go here:

http://www-01.ibm.com/support/docview.wss?uid=swg21596944

It's a Perl script that summarized deduplication by client type

 

[marclant]

Hi Daniel,

That perl script works well. This script was also written before there were any tools built-in to TSM. It's a bit more time consuming to run in TSM, but you can use "generate dedupstats" and "query dedupstats" to get similar info. The time consuming part is that you have to use "generate dedupstats" for every node individually to generate the stats.

Marc...