Surestore E 2/20 Ultrium poor performance

[Ricardo Rocha]

Hi

Does anyone have any idea why I get such different performances on the 2 Ultrium drives. Drive 1 writes 5x slower than drive 2. Drives have already been swapped, but problem remains in the drive at position 1.

Thanks,

Ricardo Rocha

[David Ruska]

Are the drives on similar interface cards? Ultrium drives are very picky on having dedicated a interface card (per drive) to maximize streaming.

Can you swap the cables to each drive, and does the problem follow the drive or the cable? That would help isolate the problem to the drive (and not a software config, interface card configuration, or cabling issue). If you do swap cables, you will need to make sure the SCSI IDs are set properly.

Are the drives on the same level of firmware? There have been a number of performance improvements in recent firmware. Sometimes replacement drives do not have the latest code.

[Vincent Farrugia]

Hello,

I assume that the 2 Ultrium drives are in a 2/20 tape library.

I think it is a firmware issue. Do the drives have the same firmware version? Check this from the control panel. From Information Menu, go to Firmware Revisions and Tape Drives.

If they do not match, upgrade the one to the same firmware level (I would assume that this is the slower drive).

HTH,
Vince

[Ricardo Rocha]

The firmware of bothe drives is 59V. And the controller is 1330. I think these ones are the latest.

Yes, both drives are in a 2/20 library, so only one fiber channel connects to it. We have several servers connected to it through a brocade. There is no overlap of the backups.

[Vincent Farrugia]

Hello,

Can you give us a diagram of how the 2/20 is connected SCSI and FC wise?

[Ricardo Rocha]

Here it goes...

[Vincent Farrugia]

Connections are correct. I assume the green dot indicates a terminator.

I really don't know what is the issue. Seems like something internal. Hope you get better answers than mine.

Vince

[Jan Klier]

If I understand your original statement correct, drive 1 and 2 have been swapped against each other, however the lower performance stayed with the drive bay 1, not with the same drive, correct?

That would indicate that the problem is not specific to one drive, which would rule out firmware, etc. Cabling would be one option, but that has already been verified, and your diagram shows correct cabling.

That only leaves other factors that remain with the drive bay.

One more possibility: Is the lower performance experienced with multiple tape cartridges? And when you swapped the drives, did the tape cartridges go with the drives, or stay in position?
Reason: a problem with a particular cartridge could result in the drive having to do more retries, which slows performance. If the cartridges stayed, that could explain it.

Also, the Device Analysis section of a HP L&TT support ticket has numerous performance related values that could provide clues. Run your back/and or test that shows the performance differential. Then generate a HP L&TT support ticket (perferably while the cartridge are still loaded in the drive) and sent it to ltt_team@hp.com. You can download HP L&TT from http://www.hp.com/support/tapetools.

[Jan Klier]

One more thing to check on the cabling: Are both terminators LVD/SE terminators? If they have the green/red LED to indicate mode, they both should indicate LVD mode. If one of them happens to be an SE only terminator, that could force the bus into SE mode, which would result in slower performance as the bus isn't allowed to operate at top speed.

[Ricardo Rocha]

Both terminators are LVD/SE. They have the green led on, I suppose it means it's running on LVD mode.

The only difference in the bus structure is that drive 1 is in the same bus as the robot, while drive 2 is "alone" on its bus. It still doesn't explains such a difference in performance....