Size of vault data prior to being CAB'd

[EVSpinner]

Hi everyone, my vault data is configured to use CABs. The size on disk is 3TB

Given the CABs are compressed data, that would mean that the DVS files if not in CABs would be more than 3TB

My question then: Is it possible to calculate how much disk space my CAB'd vault data (3TB) would use, if it were not all in CAB's?

[GertjanA]

Hello,

Normally speaking you are right.

However....

The Enterprise Vault files (*.dvs, *.dvssp and *.dvssc) are by nature already compressed. The process of adding them to a cabfile will not compress them again. You can verify by copying a cabfile to another location, and extract the cabfile there.

the process is designed to create larger files, to aid in backup performance. (many small files backup slower than fewer large files). It is not comprssing already compressed data.

You can compare it to adding many small zipped files into 1 large one.

Check this article: https://www-secure.symantec.com/connect/articles/enterprise-vault-backups-collections-and-you

And this one: https://www-secure.symantec.com/connect/articles/how-collections-and-sparse-collections-work

The last one contains also what I say above.

[EVSpinner]

ah, ok thanks for the clarification Gertjan

So is it possible then to estimate or calculate how much space I'd need if I was to unarchive a mailbox or a vault back to uncompressed format - I guess I was focussing on the CABS when in fact I should have been the DVS files?

[GertjanA · Accepted solution]

Well, this is a trick question.

I'll try to explain.

A mailbox has been archived for 10 years. You have Vault Store Partitions, having approximately 1 year of archived data. (i.e 10 partitions).

9 partitions are having cabfiles, 1 is open, and does not have cab-files. The closed (9) partitions are 500GB in size, and you have 50GB free on each partition.

If you do an export of the archive, cabfiles are copied back to the original location (with an extention archcab).

Then the cabfile is extracted, and the resulting files get also an arch in the extention (archdvs, archdvssp, archdvssc) The arch-files are written as msg in the PST file. If (if I am not mistaken) the arch files are not touched the next 24 hours, they are automatically removed. If this user has many items archived in one of the closed partitions, there is a risk freespace drops to 0, then EV-stops.

I have never seen this happen during the export of 1 archive, but there is alwasy a possibility it happens because you have little free space on the partitions, all info is on one partition etc. Having approximately 10% free space on the partitions is I believe the recommendation, and should allow an export work.

The problem you face that it will require a lot of sql checking to see in which partitions the items are stored, calculate the size, and then calculate the possible space required. I've exported many archives (also large ones, resulting in PST-files of 20GB, (or 18 PST's of 2GB) without issues. If you are unsure, you could: stop evservices, search for the *.archxxx files, remove all of them. start ev-services, export the one archive to PST. There is on the forum a query somewhere where you can find the size of the items in the archive (and you need the original size).

If I was you, I would start the export during the day, and make sure to monitor diskspace. YOu can 'granularly' export (i.e. use a date-range), but that complicates exporting.

[GertjanA]

I use this one (for EV10), but works for lower versions too:

SELECT A.ArchiveName "Archive Name",
       COUNT(S.IdTransaction) AS ArchivedItemCount,
       SUM(S.ItemSize)/1024 "Compressed Size (MB)",
       SUM(SP.OriginalSize)/1024/1024 "Uncompressed Size (MB)"
FROM   EnterpriseVaultDirectory.dbo.Archive A,
       EnterpriseVaultDirectory.dbo.Root R,
       YourVaultStore.dbo.ArchivePoint AP,
       YourVaultStore.dbo.Saveset S,
       YourVaultStore.dbo.SavesetProperty SP
WHERE  A.RootIdentity = R.RootIdentity
  AND  R.VaultEntryId = AP.ArchivePointId
  AND  AP.ArchivePointIdentity = S.ArchivePointIdentity
  AND  S.SavesetIdentity = SP.SavesetIdentity
  AND A.ArchiveName = "Fill in Archive NAme"
GROUP BY A.ArchiveName
ORDER BY A.ArchiveName

Change YourVaultStore to the name of your database, and "Fill in ArchiveNAme" needs to be the name of the archive your exporting ofcourse

[JesusWept3]

I could totally be wrong on this, but i think when you do archive exports, the arch files are removed immediately, the same way as EVSVR removes them immediately, its only user-recalls that keep the items for a couple of days or so.

But again I could be totally remembering that wrong.
But calculating how much space to extract for DVSSP and DVS files is nigh on impossible for accuracy, but the query gertjan gave above will give you a worse case scenario option, it could be a lot less

[AndrewB]

in case the follow up question was simply how much space will an archive take up when you export it, the vault store usage report will show you the original size of each archive and if you add a bit of padding for PST overhead, you'll get a pretty good idea.

[EVSpinner]

you guys are awesome, thanks very much for the info

[AndrewB]

can you take a moment to mark the post(s) that helped you as solved?