------------------------------------<snip>-------------------------------------------
My take is that most of the execution time of any given unit of work
such as a job step or CICS transaction is spent executing system code
including access methods, CICS and DB2 supplied code. An SQL statement
presumably would spend most of its time executing DB2 code. What this is
saying is that for the most part optimizing COBOL programs or REXX code
probably has relatively little payback compared to optimizations like
VSAM buffering. My take is that if a good set of guidelines is available
and used, most code and tool use will be efficient enough.
-----------------------------------<unsnip>--------------------------------------
Clark, I'll concede PART of the validity of your point. Consider the
following example:
Program before checks
BEGIN LOOP.
OPEN DB FOR UPDATE.
UPDATE DB RECORD.(s)
CLOSE DB.
END LOOP.
Run time averaged 300-330 minutes.
Program after checks and changes:
OPEN DB FOR UPDATE.
BEGIN LOOP.
UPDATE DB RECORD(s)
END LOOP.
CLOSE DB.
Run time now 20-25 minutes.
It's ok to generalize and say that most time is spent in
non-user-written code, but abuse of that code can also be incredibly
damaging. Badly written code by a shop's programming staff can lead to
terrible performance just as easily as badly written IBM or OEM code.
This example was drawn from actual experience; I can still name the
programmer, even though he's now retired.
Rick
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
