What kind of microcode is that Debbie???
Hoa.
-----Original Message-----
From: Weeks, Debbie [mailto:[EMAIL PROTECTED]]
Sent: Friday, April 27, 2001 12:35 PM
To: [EMAIL PROTECTED]
Subject: Re: LTO performance - Update
My CE just finsihed putting the latest microcode on the drives. You would
not belive the difference this made! My databse backup that had been taking
4 HOURS completed in 10 MINUTES right after the code was applied.
Perfomance improvement is an understatement!
I executed the command to write to tape again, and it completed in 8
seconds, a major improvement. We also tried writng 16M to tape for a
comparison, and it completed in 13 seconds, so some of that time must be to
locate on the tape.
Thank so much for all of your help and suggestions!
-----Original Message-----
From: Richard Sims [mailto:[EMAIL PROTECTED]]
Sent: Friday, April 20, 2001 4:01 PM
To: [EMAIL PROTECTED]
Subject: Re: LTO performance
>timex dd if=/dev/zero of=/dev/rmt1 bs=1024 count=1000
>
>The result was real 21.06 seconds. This falls way short of what I would
>expect when the advertised sustained transfer rate is supposed to be
>15MB/sec native (30MB/sec compressed). ...
Debbie - That indeed sounds poor. As Nathan suggested, look into the
microcode level. Also make sure you have a recent/latest
device driver level in the 3466.
I also wonder about the SCSI (assuming that's the 3583 attachment method).
Look for errors in the AIX Error Log. Use SMIT to check the 3466's
SCSI adapter for having correct settings. If there is a Host Interface
Board in the 358x, check for it being in the correct mode (LVD/HVD) to
match the SCSI adapter in the AIX system. Assure that the drive is
terminated, SCSI-wise, and that the correct type of terminator is used.
I presume that the 358x units are the only things on the SCSI chain,
which is to say no mongrel stuff to cause the SCSI bus speed to be
dropped to a low common denominator, and nothing to
generate SCSI noise on the bus.
Something is definitely misconfigured there. Given that it's a 3466,
perhaps IBM should be doing the analysis to find out what's wrong.
Richard Sims, BU
