Incremental backup takes forever

[Rick_DeBay]

I'm only seeing 9MB/s for an incremental backup where the target device is a disk.  That's ridiculously slow given that the network, servers, and disks are not doing anything else.

[Ken_Putnam]

What kind of throughput do you get for a FULL of the same resources?

[Rick_DeBay]

Much slower as the destination device for the full is currently a USB 1.0 device; about 1.6MB/s.
The destination for the incremental is an idle raid 5 array.

[teiva-boy]

USB 1.0 = 1.5Mb-12Mbps  That bits NOT Bytes.  So 1.6MB/s is about right.
USB2.0 I've seen up to 30MB/s when it's max theoretical is closer to 60Mb/s.  It'll never get much higher than 30MB/s in real world.

EIther way, you need to JUNK that stuff NOW.  Get something that is at least USB 2.0, eSATA, Firewire 800...  Although those all have no place for a server...  You should be using SAS, or SCSI for reliability sakes.  

A nice little boost I've found for incrementals is to NOT use the archive bit, but rather "Modified time," and the NTFS Change Journal.  But I suspect there is something like your AV settings are not set right for BackupExec, or an incorrectly configured RAID5 array (diskpart align= command, small sector/cluster sizes, no write-back cache, etc)  Or even updating BackupExec, but not updating the agents afterwards...

[Rick_DeBay]

I'm not concerned about the USB performance; I understand that USB 1.0 is slow.  Once we have a USB 2.0 connection the removable drive will use that.

The raid array is not misconfigured, BE 9.1 had a MUCH greater incremental throughput.  And all other operations on the servers do not suffer from disk performance issues.

The incrementals are using modified time.

Given all those points, I posted here as the only conclusion is that BE 2010 is using an inefficient algorithm.