If you think about how de-duplication works this question doesn’t translate well to storage where de-duplication used.

An Avamar server will contain a many unique ‘chunks’ of data.  Each unique chunk may be referenced by multiple client backups.  If any of those backups were to be removed, the chunk of data would still remain on the system until there are no other references to it.

How would it be decided which backup that chunk of data belongs to?  The original backup that uploaded the chunk may eventually expire so that chunk would have no owner.. or would it?

If you’re looking to learn about the storage demands made on the system by a particular client there is a useful script (capacity.sh) that will show you what percentage of new data the client is adding to a system on a day to day basis.  You can learn more about it with the following KB articles

esg103967 How to use capacity.sh script to analyse daily data change on the Avamar system
esg108548 How to increase the number of 'High change clients' displayed by capacity.sh

Hope that is of some use!

In fact this is also what we are looking for.

In our mutualized environnement, this would be a real help if we could find a sql query that would give us the total number of chunk a client has access too hence the total size occupied by this client on avamar even if it of course includes shared data.

If you have any idea...

Nicholas Ricioppo wrote: An Avamar server will contain a many unique ‘chunks’ of data.  Each unique chunk may be referenced by multiple client backups.  If any of those backups were to be removed, the chunk of data would still remain on the system until there are no other references to it. How would it be decided which backup that chunk of data belongs to?  The original backup that uploaded the chunk may eventually expire so that chunk would have no owner.. or would it?

Yeah thats an interesting point. In that situation you could:

1) Find the second client who referenced that chunk and is still referencing that chunk, they become the new owner and inherit the chunk. This isn't really "fair" for billing purposes. There might be another 8 clients referencing that chunk who get a free ride?

2) Forget about any ownership idea, instead work out how many clients reference that chunk and divide the cost of that chunk between all the clients who reference it. This is more fair but unrealistically complicated.

I really don't care about billing. I just want to know how much space a particular server is using as a point of reference for worst case planning purposes when adding new clients.

Hi Jason,

that would be nice how one find which chunk is reference by which client ???

This is our main question.

I'm fine doing some calculation if we can only get raw data with multiple chunk but at least i would like to know which client has access to which chunk !

I'm not sure if the product management team review these forums but if you'd like to see a new feature you can submit a request for enhancement via Powerlink explaining what you need, why you need it etc.

Go to Home > Support > Request Support > Request a Product Enhancement