On 6/6/2023 6:42 pm, Seymour J Metz wrote:
4096? That sounds like it is tuned to the underlying CI structure of zFS.
It makes no difference what the record length is. I can updtae the gist
if you suspect I'm dealing from the bottom of the deck!
Would you consider performance parameters on DCBE to be cheating?
Of course not. It won't make up for a 5x difference.
--
Shmuel (Seymour J.) Metz
http://mason.gmu.edu/~smetz3
________________________________________
From: IBM Mainframe Discussion List [[email protected]] on behalf of
David Crayford [[email protected]]
Sent: Tuesday, June 6, 2023 6:26 AM
To: [email protected]
Subject: Re: z/OS 3.1: Now UNIXR Certified
On 5/6/2023 6:07 pm, Seymour J Metz wrote:
OTOH, benchmarks are tricky things, and it is often easy to get the answers you
want by carefully cherrypicking the details. I suspect that QSAM really is
faster for the test he ran.
QSAM may have stood a chance against HFS but it has no chance against
zFS. It's a very standard benchmark which simply reads/writes 4096
length records. The code is in the public domain so I challenge anybody
to try to tweak the code so that QSAM performs better. If you can, then
leave a comment in the gist and DM me your details. I will send you a
bottle of fine claret, maybe a burgundy if that tickles your fancy. I'm
that confident it won't happen :)
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
