poor frecover performances (10hrs)

[Adegoke Apalara]

Thanks,
I checked the block size and it is already set at 8192. Attached is a screenshot of the output of fstyp -v /dev/vg01/lvol1.
What size then should i change it to.
The SCSI DLT Drive is on it's own controller and what must i do to enable LVM stripping for the alternate controller.

[A. Clay Stephenson]

I think that you need to stop and take a deep breath. Note that the backup performance is not bad -- with exactly the same hardware, filesystem block sizes, mount options, kernel tunings, .... the difference is the operation itself. Block sizes are of almost no consequence in vxfs filesystems because they are extent-based and again whatever it is it seems to be okay because the backups are okay.
Of course, the restore is blocking on disk io - what else would it be doing?

The fundamental problem is that restores require directory rewrites each time a file is added (and that causes i/o blocking) --- something that backups don't have to do.

[Michael Steele_2]

For 99% of the time LVM striping may not be needed.

Since restores are not the normal state of your server more data needs to be collected. If the problems continue with those disks then impletment LVM striping.

Restores of full backups are faster than incremental. Resize your database frequently.