I am currently backing up a mixed Unix/NT network using ARCSERVE 2000 on a central NT (SP6a) server with an exabyte M2 Mammoth tape drive attached via an Ultra 2 LVD SCSI interface. The tape drive recently started to fail on an increasing number of tapes until it would not write to any tapes. I contacted Exabyte support and they requested some log files e.t.c which I sent.
They told me the drive was indeed having problems and after some further tests they are replacing the drive. The Engineer (who was very helpful) however mentioned that our backup throughput was anything from 1.5Mb/sec to 8Mb/sec. Exabyte recommend that for these drives to run at their optimum that they be fed data at least 8mb/sec and ideally at 12mb/sec. The engineer explained that if the drive was not streaming it would be constantly stopping, rewinding and restarting while backing up. This stop start causes more wear on the tape drive and will cause it to fail sooner.
I read Exabytes documentation for the M2 tape drive and it appears this drive is designed to be used for backups over Gigabit networks or fast local SCSI interfaces and not 100BaseT switched networks like the one we have here(tape was bought by previous system admin not me!!!!).
I can only see a few methods of resolving this throughput issue:
1. Upgrade to Gigabit ethernet (costly!)
2. Place a large fast SCSI 150Gb hard disk in the backup server, pull all the backups (using tar for Unix boxes) there and then pull these directories to tape. I am not sure whether this will take longer than the current backup windows and I am not sure if ARCSERVE will support this?
3. Change the tape drive to something more suitable.
Does anyone have any other ideas or solutions that they have used to resolve this type issue?
They told me the drive was indeed having problems and after some further tests they are replacing the drive. The Engineer (who was very helpful) however mentioned that our backup throughput was anything from 1.5Mb/sec to 8Mb/sec. Exabyte recommend that for these drives to run at their optimum that they be fed data at least 8mb/sec and ideally at 12mb/sec. The engineer explained that if the drive was not streaming it would be constantly stopping, rewinding and restarting while backing up. This stop start causes more wear on the tape drive and will cause it to fail sooner.
I read Exabytes documentation for the M2 tape drive and it appears this drive is designed to be used for backups over Gigabit networks or fast local SCSI interfaces and not 100BaseT switched networks like the one we have here(tape was bought by previous system admin not me!!!!).
I can only see a few methods of resolving this throughput issue:
1. Upgrade to Gigabit ethernet (costly!)
2. Place a large fast SCSI 150Gb hard disk in the backup server, pull all the backups (using tar for Unix boxes) there and then pull these directories to tape. I am not sure whether this will take longer than the current backup windows and I am not sure if ARCSERVE will support this?
3. Change the tape drive to something more suitable.
Does anyone have any other ideas or solutions that they have used to resolve this type issue?