Sunday, September 20, 2009

De-Dup to Dupe the DBA?

We are in the process of evaluating a new SAN machine for our data center and are in contact with various vendors. Their pre-sales guys seem to be pitching the one option quite aggressively while mentioning the others just in passing and that jewel of the crown is "De-Duplication", which means that if you have identical blocks on the SAN disks, then SAN would only store one copy of it and will save you space.

In every presentation I sit through, I get hammered by these guys about this chic new feature and they spend more than half a time on this and in remaining time,they wink and refer to other features as cool but not so cool as "De-Dup".

I plan to use the half part of SAN as shared storage for my RAC, while the other part for the RMAN backups. Now I am at loss because:

1- I don't think that my database would have too many identical blocks (if any), and if they are there, then we need to seriously re-visit the database design, instead of 'De-Duping.'

2- RMAN doesn't back the empty blocks.

So, apparently, there won't be many identical blocks on my SAN.

So how De-Dup is the silver bullet for my requirements?

I am afraid to ask the pre-sales guy as they are too enthusiastic to be interrupted and besides I am in the habit of heart-breaking.


Chen Shapira said...

If you keep multiple non-incremental backups on the same volume, you will have identical duplicate blocks between the backups. De-dup will eliminate these. On the other hand, some will say that having duplications is the entire point of backups...

If you are going on the new SAN adventure you are probably planning a POC anyway. Test the De-Dup during the POC too.

Fahd Mirza said...

Chen, thanks for your valuable comments.

Unknown said...

Fahd, how did your de-dupe procject go, did you guys go for it and see any savings in space
I am curious to know
Can you post an update pls