View previous topic :: View next topic
|
Author |
Message |
nynali
New User
Joined: 08 May 2007 Posts: 13 Location: Hyderabad
|
|
|
|
Hi,
I want to know some options in SYNCSORT which will optimise the CPU usage time and hence reduce the dollar cost of the job. |
|
Back to top |
|
|
Aaru
Senior Member
Joined: 03 Jul 2007 Posts: 1287 Location: Chennai, India
|
|
|
|
Proper use of INCLUDE/OMIT (for removing the unwanted records) and INREC/OUTREC can increase the speed of the SORT thereby decreasing the CPU time.
I cannot think of any options as such. |
|
Back to top |
|
|
Frank Yaeger
DFSORT Developer
Joined: 15 Feb 2005 Posts: 7129 Location: San Jose, CA
|
|
|
|
Quote: |
I want to know some options in DFSORT which will optimise the CPU usage time |
Here are some things that might help:
- Use large block sizes for the input and output data sets
- Ensure that sufficient virtual storage is available - ensure DSA=0 is not in effect - increase the DSA=n value, if appropriate
- Use the INCLUDE or OMIT statement to remove records before sorting, if appropriate
- If you don't need the entire record for output, use INREC to shorten the records before sorting
- Use built-in DFSORT functions rather than user exits, if appropriate |
|
Back to top |
|
|
nynali
New User
Joined: 08 May 2007 Posts: 13 Location: Hyderabad
|
|
|
|
I have SORT job which does sorting and removing duplicates based on 4 keys.
There will be approximately 110,000,000 records in the input and the LRECL of the i/p file is 74 bytes.The options that were given for the SORT job are
PARM='VSCORET=40M',REGION=48M,FILSZ=E105000000
The actual FILESIZE is 8,052,395,174 BYTES.Also the sorting is taking around 22 SORTWK datasets.Going by the above options is there any way to improve the performance in terms of CPU time?? |
|
Back to top |
|
|
Frank Yaeger
DFSORT Developer
Joined: 15 Feb 2005 Posts: 7129 Location: San Jose, CA
|
|
|
|
nynali,
The way the PARM is specified, FILSZ=E105000000 will not be used since its not actually passed within the PARM parameters. I can't tell whether you actually need the FILSZ parameter based on what you've shown.
If you want me to comment further, you'll need to rerun your job with the following added:
//SORTDIAG DD DUMMY
Then send me the //SYSOUT messages offline (yaeger@us.ibm.com). Put DFSORT somewhere in your Subject line to catch my attention. |
|
Back to top |
|
|
nynali
New User
Joined: 08 May 2007 Posts: 13 Location: Hyderabad
|
|
|
|
Frank,
You are correct The FILESZ parameter is being given in the SORT CARD in the SYSIN.
Can u please let me know what more information is needed to decide upon optimization of the mentioned SORT job in last post? |
|
Back to top |
|
|
Frank Yaeger
DFSORT Developer
Joined: 15 Feb 2005 Posts: 7129 Location: San Jose, CA
|
|
|
|
I received the information you sent me offline. However, the WER messages indicate you're using Syncsort, not DFSORT. I'm a DFSORT developer. DFSORT and Syncsort are competitive products. I'm happy to answer questions on DFSORT and DFSORT's ICETOOL, but I don't answer questions on Syncsort.
I'm moving this thread to the JCL Forum. |
|
Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006 Posts: 19244 Location: Inside the Matrix
|
|
|
|
Hello,
How large was the information you sent to Frank?
If it is not large and you post the the information you sent to Frank offline, someone here may be able to help. |
|
Back to top |
|
|
Alissa Margulies
SYNCSORT Support
Joined: 25 Jul 2007 Posts: 496 Location: USA
|
|
|
|
nynali,
Feel free to email me your complete job listing as a text attachment to alissa.margulies@syncsort.com and I'd be happy to help you tune your application.
Regards, |
|
Back to top |
|
|
|