Question on multiple staging servers performance

Being familiar with Oklier's ATO configuration, here's a few thoughts on the situation:

1. He has all clients from Avamar in a single ATO group, absolutely the wrong thing to do because all clients will be processed at one time and their data  recovered prior to initiating any tape BU.

2. A given ATO client group as documented should be sized to accommodate at least the ATO recovery phase within a 10-12 Hr window after which the tape BU would be initiated. the various client groups could then be scheduled on different days throughput the month in a managable way.

3. If you have clients with large FS containing large file counts these should be configured to use ATO's incremental staging. The use of incremental staging does not impact whether the tape BU is full or incremental because a full copy of the clients data will alway be present on the staging disk. What incremental leverages is the fact that most FS BU's contain many files never change from one backup to another therefore why waste the time to stage them each and every month? Incremental staging can shorten the stgaing phase often by a factor of several times.

4. Oklier's tape application is BackupExec which does not support tape multiplexing therefore only a single stream can be backed up to tape at a time.

5. Multiple staging servers will benefit all ATO configurations because what it facilitates is the ability to perform multiple concurrent staging activities therefore you can accomplish twice the workload with two staging servers as you can with one. ATO can support up to 20. If your recovery speeds from Avamar are poor then multiple staging servers will also be beneficial because two slow sessions are usually better than one.

6. Using a virtual system as a staging server is acceptable but it almost never performs as well as a physical server. Also, it is not possible to connect a tape drive to a VM system therefore the tapeout data will have a minimum of two network hops in order to reach the storage node or media server hosting the tape drive. Ideally a ATO staging server should be a media or storage node server as it eliminates the need for the second network hop.

Well the problem is

4. Oklier's tape application is BackupExec which does not support tape  multiplexing therefore only a single stream can be backed up to tape at a  time.

Also what i have done is look at my major backups or tape outs. I had 3 file servers that were a few TB in size. I have excluded these from ATO and set a job in Networker to only back these up on the 2nd of the month. This way i don't have to stage these and hold the duplicated data topping out at 9tb.

For my one staging server that handles ATO and Networker clients. I have bonded all 4 Nics to switch assts load balance my data. If you wanted to use your Storage Node with SAN/NAS disk then you could present some disk to this client and stage more to a different Drive/Dirctory.

How many tape drives do you have?

Thank you both for responding. The time I was looking to shorten was the stage out. Yes backupexec can only handle 1 stream, but since I am going to LTO4, I can tape out at around 3.5GB/min (what the software reports). I think I'm going to go with the Incremental tape out first before I try multiple servers. The other thing I am going to try is to use multiple datastores just because I dont think I have enough on the single array. Since our tape outs are historical only, we are goign to skip servers that dont need to be staged out, i.e. every DC, so we'll just stage out 1. This should minimize the amount of data that gets rehydrated.

I'll post again when I have done two incrementals, since the first will be no time savings.

Thanks again!