Backup Exec running slowly, only writing every 30 to 40 seconds

Question

We have an Idealstor BA4U2E server and a 2000-TLYTE-2RSA (2-bay SATA rack enclosure) connected to it. We run Symantec Backup Exec 12.5 on it to copy between drives.

Lately our copy operation has been running significantly slower than it should. We also noticed, by remote desktopping into the server, that it has pauses for around 4-5 seconds, every 40 seconds or so. During these pauses everything in the system completely freezes. Monitoring the system with Process Explorer, we found that bengine.exe (the Backup Exec engine which does the copying) idles at 0 cpu for the 40 seconds, and then the whole system does it freeze, and then there's a huge I/O and CPU spike in the graph, during which bengine.exe was reading and writing to the disk. Monitoring even closer with Process Monitor does indeed find the 40-second gaps in which bengine.exe is doing nothing.

What is causing our copy operation to do this? How can we diagnose the problem?

Answer 1

We just got it back up to fast speeds, but can't replicate it. Here are the things we did before realizing speeds were faster:

• Stopped Trend Micro antivirus; but, it was already set to not scan the drives being backed up from and to, and fast speeds have been achieved in the past with Trend Micro running, so this should not be the issue.
• Ejected one of the hard drives. A custom drive management application is used to eject the drive, which is provided by Idealstor, and its service was not running so we had to start the service, then launch the program, and then use the program's eject function.
• Replaced that hard drive into the same slot. (this was the left SATA drive on the 2-bay enclosure)

That's all that was done. We think it could be some anomaly that was magically fixed by pulling the drive out and putting it back in. But restarting the antivirus (and restarting the task) didn't slow the task, nor did stopping the management service and restarting the task, or exiting the management application and restarting the task. We could not slow the task back down, so we can't trace the cause.

Also just a note, it is not a temperature-related issue; the reason for pulling out the hard drive was to test its temperature, which was found to be a cool 27 C, even after the backup task had been running for an hour beforehand (albeit very slowly, as explained). The individual hard drive enclosures each have two small fans at the front of them, and our server closet has a constant cold airflow, so there is absolutely no chance that the drives or the controllers are overheating.