Exchange 2010 backup questions

[ollie1970]

Dear All,

We use Exchange 2010 Enterprise on vSphere 5.1. We currently have 1 mailbox database of 1,7 TB on a 2 TB lun. We use Veeam Backup & Replication v6.5 for backups. Because of the snapshot Veeam creates, the LUN becomes full and Exchange stops.
I want to move al mailboxes to other Exchange 2010 servers with new to make databases and compress the original database. All Exchange servers are part of DAG's. Can I compress a database in a DAG? What is the recommended database size for Exchange 2010 with a LUN size of 2 TB & Veeam backup with snapshotting in mind?
Is there a maximum of databases per Exchange 2010 server Veeam can back-up?

Any help is highly appreciated

[Vitaliy S.]

Hi Olivier,

What is the recommended database size for Exchange 2010 with a LUN size of 2 TB & Veeam backup with snapshotting in mind?

I would recommend to have at least 10-20% of free space for the Exchange VM snapshot.

Is there a maximum of databases per Exchange 2010 server Veeam can back-up?

There is no such limit.

Thanks!

[ollie1970]

Thanks Vitaliy S.

As for Exchange storagedesign with Veeam back-up in mind, what is preferred; 1 LUN of 1 TB with 2 vmdk's of 400 GB or 2 LUNs of 500 Gb with 1 400 GB vmdk each?

Many thanks in advance!

[Vitaliy S.]

Hmm....I don't believe there will be a big difference, but having Exchange virtual disks on different datastores might be a better design cause of the less total I/O load during snapshot commit operation.

[dellock6]

... as long as those two datastore belong to different disk pool/raid groups, otherwise I/O will hit at the end the same disks.
Going a little bit offtopit to Veeam and into virtual machine design, one large lun is usually preferred nowadays because it allows for bigger free space and disk allocation.
In your design, starting from 2 vmdks each large 400 Gb, with 1TB single LUN you have 200 Gb of free space to allow future growth, with 2 separate lun each vmdk can only grow for other 100 Gb. Free space is not only needed for snapshots, but also for other files like swap...

Luca.

[Vitaliy S.]

... as long as those two datastore belong to different disk pool/raid groups, otherwise I/O will hit at the end the same disks.

Good point, Luca!