Tape device I/O statistics

[James Morrish_1]

Hi,

Does anyone know how to acquire tape device throughput statistics within HP-UX? I have Measureware/Glance installed on the system.

Apparently this can be done with the iostat command in Solaris, but there are few options for the HP-UX version.

I am trying to identify faults with our Netbackup SAN environment, and would like to verify the data throughput.

Environment:
- HP-UX 11i v1
- IBM atdd driver v3.5.0.20
- IBM Ultrium LTO2 3580 tape drives (SAN attached)
- Netbackup 5.1 MP4

Thanks in advance,
James

[A. Clay Stephenson · Accepted solution]

The best numbers I have found are obtained by Glance.

Processes -> Select a tape drive related process

Next, Reports ->Process Open Files

You should then note how the File Offset for the tape device node changes over time.

[Steven E. Protter]

Sorry so slow, but ITRC is quite slow now.

I think glance or sar tools, with glance being the better choice will do the job for you.

Don't expect the actual throughput to reach anywhere near the theoretical speed of your scsi connection.

Real life won't appraoch that.

SEP

[Alzhy]

Greetings!

Peace & Clean Earth!

I do not think there are no tape device measurement tools available on HP-UX...

SEP, sar only does that on Solaris.

FWIW.

[James Morrish_1]

Thank you for your fast response. The offset against the tape device in the glance output looks just the ticket! I'm looking at how to capture this data to a file (cron/awk maybe).

I've defined the Netbackup bptm process as an application within Measureware, but it only appears to record the reads from disk. All other I/O appears to be '0' or a negligible amount.

Thanks,
James

[Nick W]

Hi James,

I agree that Glance is probably going to be your best bet to get 'live numbers'

I suggest you also look at the Performance Assessment Tools (PAT) and Lib and Tape Tools (L&TT).

There is even a 'howto' document with links to the tools:

www.hp.com/support/pat

These should help you to identify where in your setup the bottleneck may be happening.

Do Netbackup Support or IBM support folks have any suggestions or tools which may also help you?


Regards,
Nick

[James Morrish_1]

Hi Nick,

Thanks for the information. Funnily enough, someone has forwarded me the link to the PAT site today.

I have had problems with vendors accepting responsibility due to the multi-vendor configuration of this solution, but will persevere.

It looks like our storage team may have identified a fault on the SAN which is causing the HP systems to have problems. When the link with a suspected problem is active Netbackup slow-downs and backup hangs are experienced.

Thanks,
James