Time for collection enable

[MontoN]

Dear All,

We need to enable collections on EV partition. We have 4 partitons out of which one is open. The total Ev data is around 7 TB. Can any one inform me that , thow much total time EV will take to enable collections on EV data.

And most importantly, do we require backup of EV before enabling EV collection for 7 TB of data?

Thanks,

MontoN,

[JesusWept3 · Accepted solution]

Whats the reason for needing collections?
Typically they're more trouble than their worth, their counts get out of whack, takes ages to reclaim diskspace due to sparse collections and not being able to delete items out of cab files, also can take up a lot of space when it extracts items to archDVS Files, makes processes like index rebuilds, PST Exports etc 25% slower, and also theres a locking issue with EV10 and below where items can't be extracted etc

Also CAB files don't save any diskspace, since the EV collections are uncompressed.

If its for backup speed you may want to look at Flash backup with netbackup or a snapshot type of deal depending on the device

But regardless, I honestly couldn't tell you the speed, 7TB is quite a lot and depending how spread that is it could be quick or it could be slow.

For instance if you have 12 servers with 580GB each, that would be a lot quicker than 2 servers with 3.5TB each. But honestly I've never seen any metrics regarding it, i think initially it will be quite slow, because it will be going through each file and folder on the server and that will take enough time as it is, and it will be hammering away at SQL as well, so if your indexes are fragmented then it could mean that the bottleneck might be SQL and not the disks etc

Your best bet if you can is maybe creating a test lab environment, archiving a few million items and then running collections and just see how that works and use that as a best case scenario, since the lab won't be actively used by archiving, searches, retrievals etc

[AndrewB]

hi MontoN, there are too many factors involved to tell you how long it will take to collect all your vault store data into CABs. however, to answer your question about backups, yes, you should certainly have regular backups of your system and especially before implementing any major configuration changes.

if you enable collections, you can start with one of the closed partitions and monitor the event logs to see the progress; you dont have to do all of them at once. also, make sure your collections schedule doesnt overlap with other operations that are going on in your EV environment.

...and p.s., i support JW in not recommending collections because there are too many drawbacks and only one benefit.

[WiTSend]

Whether or not to have collections is a debatable issue with many pros and cons, as you have see above.  I will shortly be enabling collections since my device doesn't really like have hundreds of millions of small files.  You will regain some space depending on your storage sector size.  You will definitely see an improvement in any type of backup/replication, due to the smaller number of files.

It looks like you are running about 2TB per partition.  Make sure you have sufficient space available for large re-expansion of the files during some operations.

[MontoN]

Dear All,

Thanks for the all your replies.

Regards,

Omkar Nalawade.