Removing Powerpath didn't actually help a lot in our situation. However, last weekend, we completely formatted the SAN array we're using for our backups (7 disk 500 GB RAID-5 setup) Firstly, we duplicated all our backup files to tape. Then, in BackupExec, we deleted all the backup to disk folders and files, and set our jobs to backup to tape (didn't start the jobs though, we put them on hold). After this, through Windows Explorer, we formatted the logical disk associated with our SAN. A reboot just to make sure of the server, than back into BackupExec to create the backup to disk folders. We used to est a maximum size for our backup files, but this time we've also put a mark at "Allocate the maximum size for backup to disk files"
Before the actions described above, when running the windows defragmentation tool on the logical disk, it would display an almost full red bar, only leaving out a few white spots to represent available disk space. This meant the logical disk was heavily fragmented (please note that defragmentation using Windows on a SAN can easily degrade performance in stead of improve it, depending on type of SAN, type of array and other factors. See this link for more information:
http://www.las-solanas.com/storage_virtualization/san_volume_defragmentation.php ) After this weekend, because we've put a check mark at "Allocate the maximum size for backup to disk files", the defragmentation tool shows a perfect blue bar, indicating a completely defragmented drive.
And the results... great! We went up from a maximum of 400 MB/s for two or three concurrent jobs to a maximum of 1.5 GB/s for two concurrent jobs and even experienced a max. of 2 GB/s for three concurrent jobs! Also, duplicating to tape went up from about 350 MB/s to 4 GB/s! Job averages went up from around 250 MB/s to 750 MB/s and higher.
We'll run without any additional changes this week to see wether the current speeds hold up. Next week, we'll try dusty's solution to implement Powerpath 1.1 including the fix.