Slow duplication of Exchange DAG backup images to tape.

[nairaj]

We use netbackup vault jobs to create the second copy of our backup images.

The vault duplication jobs for the windows and unix clients gives us around 250Gb/hr to LTO4 tapes but at the same time it only gives 25-50Gb/hr for the Exchange DAG images.

Any thoughts or if you can direct me to something.

thanks.

[Mark_Solutions]

Are you doing GRT Exchange backups? If so the exchange database image gets mounted and processed before it even starts writing to tape (which can tie a tape drive up for quite some time) This can be imporved by using a smaller fragement size for the disk storage unit that does the GRT backups - i usually use 5000MB for GRT backups It can also help to have the media server as the GRT Proxy Restore host so that the image is mounted on the media server (it does this when it comes to duplicate the GRT information - exchange servers client host properties, exchange section) The other option is to not duplicate the GRT information - but if you deselect that you cannot recover individual messages from tape copies in the future (Master Servers host properties) You will see if it is GRT causing the issue as you will see in the details of the job the "Processing" lines with the message numbers - until it has finished all of this it doesn't actually write to tape Hope this helps

[nairaj]

Thanks for your reply.

We don't use GRT option. Exchange DAG Database backups are taken to DataDomain first and then the images are duplicated using vault. 

[Mark_Solutions]

Dont know then - only other thought is that they are large single file backups with no check points so perhaps you could reduce the fragement size used by your DD disk storage unit to see if that helps with the data positioning within teh de-dupe and give better rehydration

Maybe try the 5000MB fragment size anyway and see what happens after the next backup / duplication