Peter,

John's initial guess was pretty much on target.  You first sort is going to do 
much more work than the second one.  You might ask why since it is sorting on 
fewer key bytes?

The answer is they share the first 3 bytes, which might not seem like much, but 
it at least groups the data.  Let's assume it is a department code, or a 
country code, it doesn't matter, all of each group are together.

Without seeing the specific job output it is tough to give specifics because 
the size of the records and the size of the keys as a percentage of the record 
length and the amount of memory all factor into how we process data.  (Was that 
enough of a disclaimer???)

Anyway, if all we have to do is do minor re-arrangements on the second pass, 
then the CPU used is much less.

One other thing, all modern sorts re-order and sometimes convert the records so 
that the keys to be sorted can be processed with a single CLC instruction.  
Therefore the first sort had to do that conversion on input and then convert it 
back to the original format on output.  The second sort didn't have to do any 
of that conversion.

If you want a fuller explanation, you need to send me offline all the sort 
messages for both steps.  Or, you can call our support folks at 201-930-8260.  
Actually, I am supposed to refer you there first, but I like to talk to real 
users directly sometimes.

<Advertisement>
On our web site at: 
http://www.syncsort.com/en/Data-Integration/Solutions/Mainframe-Solutions/zIIP-Offload-for-Copy
We have information on a new product, MFX ZPCopy.  It can significantly reduce 
CPU cycles used for copy operations.
If you are interested, please contact the support people at 201-930-8260.
<End-Avertisment>

Chris Blaicher
Principal Software Engineer, Software Development
Syncsort Incorporated
50 Tice Boulevard, Woodcliff Lake, NJ 07677
P: 201-930-8260  |  M: 512-627-3803
E: [email protected]


-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf 
Of Farley, Peter x23353
Sent: Friday, August 30, 2013 5:22 PM
To: [email protected]
Subject: Re: Peculiar CPU behavior of SYNCSORT?

You're right, the sort orders are not the same.  For the CPU example I gave 
below, the first sort uses these parameters:

$ORTPARM:       CORE=MAX
SYSIN:  SORT FIELDS=(1,3,CH,A,17,7,CH,A)

The second one gets these:

$ORTPARM:       CORE=MAX
SYSIN:  SORT FIELDS=(1,24,CH,A)

The application in question is not one I am intimately familiar with so I can't 
say offhand whether that first sort is very disruptive of the original order or 
not.

Peter

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf 
Of Charles Mills
Sent: Friday, August 30, 2013 6:10 PM
To: [email protected]
Subject: Re: Peculiar CPU behavior of SYNCSORT?

Are the keys the same? Is there a difference in how out-of-order each file is 
going into the sort?

> There is no difference in the parameters given via PARM and/or SYSIN

Must be a different sort key, right? Or what's the point? Is each sort sequence 
the same? If you sort a file into a particular sequence, make some fairly 
slight changes, and then sort it again into the same sequence -- the second 
sort is going to use a heck of a lot less resources.

Charles

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf 
Of Farley, Peter x23353
Sent: Friday, August 30, 2013 3:00 PM
To: [email protected]
Subject: Peculiar CPU behavior of SYNCSORT?

While analyzing high CPU consuming jobs we came across a peculiar behavior of 
SYNCSORT. Certain SYNCSORT steps in some of the jobs are *regularly* (not just 
occasionally) taking much more time than some of the other SYNCSORT steps in 
the same jobs with same amount of data.  For example, in one job a file is 
sorted in certain order and after enrichment by an application program is 
sorted back into the original order. The first SYNCSORT takes 96 seconds of CPU 
time while the second SYNCSORT takes only 8 seconds. There is no difference in 
the parameters given via PARM and/or SYSIN. This kind of processing is done 
multiple times in many application jobs, so if there is anything we could do to 
reduce the time taken by the first SYNCSORT we would be very interested in 
doing whatever it takes.

Can anyone suggest why there should be such a disparity in observed CPU timings 
for SYNCSORT?  Or what SYNCSORT parameters might be worth investigating to 
reduce the initial SYNCSORT CPU time?

TIA for any helping to cure my ignorance.

Peter



Example SYNCSORT CPU timings

DATE : 08/15/2013

STEP NAME   RECORD COUNT   CPU TIME
SORT10      6547982        01 MIN  28.64 SEC
SORT20      6547981        05.35 SEC

DATE : 08/16/2013

STEP NAME   RECORD COUNT   CPU TIME
SORT10      6551104        01 MIN  36.83 SEC
SORT20      6551103        08.02 SEC

--

This message and any attachments are intended only for the use of the addressee 
and may contain information that is privileged and confidential. If the reader 
of the message is not the intended recipient or an authorized representative of 
the intended recipient, you are hereby notified that any dissemination of this 
communication is strictly prohibited. If you have received this communication 
in error, please notify us immediately by e-mail and delete the message and any 
attachments from your system.

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions, send email to 
[email protected] with the message: INFO IBM-MAIN



ATTENTION: -----

The information contained in this message (including any files transmitted with 
this message) may contain proprietary, trade secret or other  confidential 
and/or legally privileged information. Any pricing information contained in 
this message or in any files transmitted with this message is always 
confidential and cannot be shared with any third parties without prior written 
approval from Syncsort. This message is intended to be read only by the 
individual or entity to whom it is addressed or by their designee. If the 
reader of this message is not the intended recipient, you are on notice that 
any use, disclosure, copying or distribution of this message, in any form, is 
strictly prohibited. If you have received this message in error, please 
immediately notify the sender and/or Syncsort and destroy all copies of this 
message in your possession, custody or control.

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to