How much data is compressed in a library

BWijnhout

Newcomer
Joined
Aug 28, 2008
Messages
3
Reaction score
0
Points
0
Hello all,

We have got two IBM 3584 tape libraries with 900 TB backup data. We want to move this tape data to Data Domian deduplication gateways and XIV disks. Before we are going to move this data we want know how much data of the 900 TB is compressed by the clients, because compressed data cannot be deduplicated.

How can we do this?

Kind regards,
Bert.
 
Short answer: no way of knowing in a very short practical way.

You can look at the node and activity logs since the beginning and work your way but do you keep logs this long?
 
Hi,
What is your question?:
how much data is compressed by the clients (B/A, TDP client)?

OR

how much data is compressed in a library? (hw compression by the drives)

Rudy
 
The data domain process works pre-tape compression so the question is easy to answer. The easy part is that TSM is not compressing the data unless specificly stated in the client options/sys file. Now if you are using a Windows client that has compressed folders I have found that the OS uncompresses the data on the fly so TSM can read it. So in most cases the data from the server to the TSM server is not being compressed unless you set it on the client or set/force it from a client options set.
 
Hi,

The question is how much data is compressed by the clients (B/A, TDP client). We keep the actlog for 40 days, so we already know how much clients use compression (about 1300 of 3000) and we know something about the ratio, but this is only about incremental backups, there a lot of active files that are not backed up daily. We want to know about the total amount of compressed data in this 900 TB.

Kind regards,
Bert.
 
I checked the tables and can't find anything that would allow you to compute how much of the data is compressed in your 900TB. So you are using client side compression? I tend to let the tape drives handle compression since they do it through hardware and there is no impact on the clients performance.
 
Hmmm, ok, I was hoping there would be a nice SQL query on the database, now we have to do a lot of investigating and calculating.
Thanks everybody.
 
So, why are you concerned with the amount of hardware/software compression that's going on if you are moving to DataDomain. (You should only use client-side compression if you have a slow or saturated LAN/WAN link.) I would think that you would be more concerned with the amount of compression you believe you will get on the DataDomain side. If you are backing up mainly file systems, the answer would be 'not much', since TSM using progressive incremental. You're only going to see significan deduplication if you do a lot of full database backups, and some cases of VMWare backup through VCB.
 
Back
Top