We are trying to calculate the storage used by some NDMP backed up images in our Netbackup 5240 appliance, in order to migrate them to another appliance and expire them in the original appliance. So we need to know how many storage would be free expiring this images.
I know we can use bpimagelist to show how many data is being backed up, but where can I see how many storage in MSDP is used by images? is there any way to know that?
This question is actually either:
a) answerable, but only if no other clients have been backed-up to MSDP, and you want to delete eveything for that one client - i.e. the answer is therefore everything
b) impossible to answer if any other clients have been backed-up to MSDP - and you no longer have *all* backup job details in activity monitor (or all bperror job records) for all currently retained backup images within MSDP
c) it is possible to approximate an answer but only if you do still have all activity monitor records (or all bperror job records) for all backups currently retained within MSDP - i.e. if you have all records of all achieved dedupe rates, because if you do have all this then you can work backwards and use the backup size versus the dedupe rate to detemine the quantity of backup data that did dedupe and therefore the quantity of backup data that did not dedupe - then you can determine how much space might be reclaimed - but you will never know whether this is a best case answer, or worst case answer, because one can never know how much of a client's backup data deduped against which other clients - i.e. you know that some backup data did dedupe against either itself and/or other clients, but which? Was it itself, or other clients, and which clients? And this complexity is further compunded if you attempt to consider "from when".
Do you see how it is not possible to answer from NetBackup, or indeed from any other dedupe application when based upon nothing but original backup size. The loss of visibility of possibly "quantifiably reclaimable space by time" is a natural side effect of dedupe. IMO, this loss of specific clarity is inherent in the nature of dedupe.
Veritas may have some private tools to inspect the "segment use counts" of the contents of their proprietary "segment containers" at the lower levels within MSDP, and from these it might be possible to total-up how much backup data stored is unique for any given client (i.e. how many segments are referenced only once and their total size) - as this would be the max reclaimable / returned space by deleting all backups pertaining to that client. Other vendors may have tools readily available for this, but I cannot confirm.
This question is made even harder if you were to also attempt to consider time frames, i.e. only select specific backup images from select specific dates or a range of dates.
I'd love to answer this question too... so, if anyone knows of a reliable CLI way to:
1) list all segments (ID or hash), segment size, segment reference count
2) list backup images and their segments (ID or hash)
...then I'm willing to have a go at scripting a re-usable utility / tool which will accept switches like client, policy, type, size, dates etc and munge the meta-data to generate reports of dedupe data sizes and unqiue data sizes.
Can anyone help getting the meta-data related to segments?