Martin Kline wrote:
If it ain't broke don't fix it? That would introduce new
potentiality for errors and would increase the overhead
for everything other than IEFBR14. How expensive is one
ATTACH and how much are you willing to pay to get rid
of it?
Oh boo hoo! If you're afraid of the -potential- for errors,
you're in the wrong business.
The extra overhead for non-IEFBR14 steps is two instructions,
CLC and BC. Since the cost of setting up the step, attaching
the program, generating stats, etc is likely many thousands
of instructions, it's a simple tradeoff.
Suppose the cost is 1/1000 of one cent per ATTACH and
ignore the other associated costs like SMF data. I'd
estimate we use IEFBR14 10,000 times per day. That's $36.50
per year. Maybe $36.60 for leap years. Do that for 30 years
for $1095.70. Do that for 1000 data centers for $1,095,700.
We don't use 10000 IEFBR14's daily, I'm pretty sure of that.
Our programmers are smart enough to code DISP=(,CATLG) just in step when
the dataset will be opened. Nevermind, let's back to the economy:
where can I get my (or my company's) $36.50 ? Oh, we're open for several
years!
But seriously: this way of cost counting is completely false. It is
obvious, that cost of CPU cycle is different during businness hours and
after hours, it is obvious that IEFBR14 will be invoked with lower
priority than online transaction, it is obvious that all the time when
CPU is not 100% busy it mean some CPU cycles are LOST.
IMHO this change is one of the least important in JCL, batch processing,
z/OS and whole mainframe industry. What would be less important ?
Just my $0.02
--
Radoslaw Skorupka
Lodz, Poland
----------------------------------------------------------------------
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
