BIAS paramter in SYNCSORT

Vidhya Kalyanasundaram · New User Joined: 19 Jul 2007 Posts: 30 Location: chennai

We are aware that Syncsort displays the BIAS option in the SYSOUT when we execute a SORT step which shows the degree of similarity of records between the Input and Output file.

If BIAS=99 - Shows that the input file is already in sorted order of keys and not much SORT process being carried out.
If BIAS=00 - Shows that the input file was entirely sorted to produce the output file meaning much difference exists between i/p and op files.

If SORT FIELDS=COPY is given, BIAS parameter is not displayed in SYSOUT. (which may mean that file is already sorted) - Correct me if i am wrong !

I executed a SORT step with SORT FIELDS=(keys). The i/p file had around 500 Records and the SYSOUT did not show me the BIAS parameter. Is there any reason for it?

Could anyone please let me know why the BIAS was not displayed in SYSOUT.

Vidhya
"If you have knowledge, let others light their candles with it. "

PeterHolland · Posted: Fri Jan 29, 2010 1:09 pm

The BIAS value is not included
in the message for an incore or turnaround sort,
where it is 100 by definition.

enrico-sorichetti · Posted: Fri Jan 29, 2010 1:33 pm

Vidhya Kalyanasundaram · New User Joined: 19 Jul 2007 Posts: 30 Location: chennai

Could you please explain 'incore or turnaround sort'?
Does it mean the i/p file is already sorted in the order of keys?

To conclude that, i tried with the below 2 SORT steps :
1. Ran first step with SORT FIELDS=(1,14,CH,A) and created the output file FILE1 - The step for which it displayed 'WER177I TURNAROUND SORT PERFORMED'
2. Ran the second step with same i/p file of Step.1 coding, SORT FIELDS=COPY and created the output file FILE2
3. Compared the 2 output files FILE1 and FILE2.

The compare results showed differences which implies the i/p file is not in sorted order of keys.

Could you please let me know when the below message will appear in the SYSOUT -

enrico-sorichetti · Posted: Fri Jan 29, 2010 4:36 pm

PeterHolland · Posted: Fri Jan 29, 2010 4:58 pm

WER177I TURNAROUND SORT PERFORMED
EXPLANATION: SyncSort was able to sort the input file without using
intermediate storage (SORTWKnn's). All input data was contained in virtual
storage.

All previous comments are stating info that YOU the TS could have found
yourself in the manual.

This is a waste of my time which i never get back.

Anuj Dhawan · Posted: Fri Jan 29, 2010 6:01 pm

Why are you so bothered for a message which is just for some information(WER177I) - I, as a Software Engineer, was always bothered about errors ...

And this

enrico-sorichetti · Posted: Fri Jan 29, 2010 6:45 pm

what should bother is that the ts does not grasp the difference between
SORT FIELDS=COPY and
SORT FIELDS=(<sort fields description>)

dick scherrer · Posted: Fri Jan 29, 2010 8:54 pm

Hello Vidhya,

Do you have the answer(s) you were looking for?

If not, please clarify what is still needed.

Vidhya Kalyanasundaram · New User Joined: 19 Jul 2007 Posts: 30 Location: chennai

Hi all,

I just gave an example to explain the 2 scenarios.

I am aware that SORT FIELDS=COPY will just make a copy of the file, without knowing it we could not code a SORT step by the way.

I had a requirement in which the incoming file has around 84924926 records(huge volume). Sorting this file on keys consumes around 4 mins of CPU. In this step only, i am not getting the BIAS keyword but getting the message, 'TURAROUND SORT PERFORMED' in the SYSOUT.

The next PROGRAM step in the JCL needs the file to be sorted with respect to the same keys. The incoming file comes as SORTED on KEYS most of the time.

So,
If we could predict that file is already sorted with any of the parameter(if through BIAS option) before the SORT process
starts,
'Whether we would be able to eliminate this SORT step?
If not, to perform the SORT step?' was my query.

This query was just to know if any tuning technique could be adopted to eliminate that redundant SORT step thereby reducing the CPU cost and Elapsed time savings.

But reply from enrico-sorichetti made me to understood that "sort will find out only after having read up to the last record"

Thanks for all your comments !!!

Vidhya
"If you have knowledge, let others light their candles with it. "

dick scherrer · Posted: Mon Feb 01, 2010 8:25 pm

Hello,

But the good news is that if the file is almost completely in sequence, it should not require the same amount of cpu as sorting a file that is almost completely out of sequence. . .