Accelerator enabled policy which needs a scheduled backup to tape once a month

[Dangerous_Dan]

Hi,

I'd like to make great use of Accelerator but have stumbled at the first hurdle.

Our business requires that we must take an end of month backup to tape which is retained for 1 year. We don't have a problem doing this until we tick "use Accelerator".

So far I have worked out that you can configure a SLP to write the accelerated image to a dedupe appliance and then duplicate that image to tape. BUT we do cross site backups and so the process of inflating the image back across site to then send it back across SAN seems clumbsy and a waste of badwidth and time.

An alternative would be to add all these clients into another policy for a tape backup.

I have also found a dodgy way around this by pointing the monthly schedule to a SLP with dedupe and saving the policy then changing the SLP to TAPE. I've tested the backup and sure enough it shows "31/08/2012 15:04:39 - Info bpbkar(pid=6648) NOT using change journal data for <C:\>: not supported with regular full backups" which is what I want but the downside is that every time you want to add a client to this policy you have to revert the SLP to Dedupe.

This is something that I feel has been clearly missed which would otherwise be perfect.

[D_Thomas]

Have you looked at using Vault to create the monthly tape? You should be able to specify where you get the data from and change the retention of the first full backup (or whatever your choose) to copy to tape.

[Dangerous_Dan]

Hi,

Yes we have looked at this but this is the same as using an SLP to write the image to Appliance first and then use vault/SLP to duplicate the image somewhere else. Why cant we just write this image to tape in the first place?

[D_Thomas]

You could create a separately policy for the Monthly job since its a full and have that go straight to tape.

I am not sure how that would affect the accelerator data on the client.

Using vault or the SLP to make a copy of the full once a month is good in some ways, its going to be faster (at least it should be) than going straight from the client to tape since the the data is read pretty quickly from the dedupe device you are using. I am often surprised at how fast the MSDP to tape transfer is in our jobs - it often soaks the entire 1Gb network on the media server with the tape drives.
DT

[Dangerous_Dan]

I don't believe the journal will be affected with doing this but the problem with this approach is that we then have to manage additional policies for 1 client which can be a nightmare to manage.

And with the second option, we do cross site backups so if we go down the SLP route we would waste lots of bandwidth:

1.Send backup data across to site B to dedupe appliance.

SLP process:

2.Hyrdate backup data and send back across to site A to media server

3.Send data back across site B to tape drives.

Seems like an awful waste.

a simple "Deactivate Accelerator" tick box in the schedule would fix this issue.