How many Cylinders or Tracks of spaces would be required

Elixir · Posted: Tue Nov 01, 2011 12:29 am

Hello,

How many Cylinders or Tracks of space would be required to write 400000000 records to a flat file each record having LRECL 133 FB?

enrico-sorichetti · Posted: Tue Nov 01, 2011 12:38 am

you were already provided suggestions on the same issue here!
www.ibmmainframes.com/viewtopic.php?t=56569&highlight=

Bill Woodger · Posted: Tue Nov 01, 2011 12:38 am

Didn't this help when it was mentioned before?

enrico-sorichetti · Posted: Tue Nov 01, 2011 12:41 am

most probably the TS wants us to provide the number without doing any work

Akatsukami · Posted: Tue Nov 01, 2011 12:42 am

Robert Sample · Posted: Tue Nov 01, 2011 12:42 am

Use the formula for half-track blocking:

1. Divide 27998 by 133 (the record length). Discard any fractional part. This gives records per block.
2. Multiply the integer derived in #1 by 30 (2 blocks / track times 15 tracks / cylinder). This gives records per cylinder.
3. Divide your record count by the result of #2. If there is a fraction, add 1 to the integer part and drop the fraction (do this even if .001 -- you're not rounding). This gives total cylinders required.

You can then adjust, if needed, by the maximum number of cylinders per volume (which will depend upon your site and which release of z/OS your site is running and whether or not the maximum 65,535 tracks per volume cylinder is in effect).

Bill Woodger · Posted: Tue Nov 01, 2011 1:12 am

I have to say, it sure is a lot.

133? Print? Who's going to read all that? That's somewhere around, give or take a few, eighty million pages. Archival? Going off-site for preparation? Why not whack it straight to tape?

If it is for something local, cut the stuff down in some way. If you can reduce the length by 20% you get a couple of drives freed-up. Got any blank lines in that lot?

OK, tell us what it is for, maybe split the savings with us?

Ed Goodman · Active Member Joined: 08 Jun 2011 Posts: 556 Location: USA

Fire up your old QuickBasic from MSDos and run this program:

Elixir · Posted: Wed Nov 02, 2011 12:24 am

I am planning for a single time execution. This file will help me in debugging the code if there is any issue with it.

dick scherrer · Posted: Wed Nov 02, 2011 12:45 am

Hello,

Pedro · Posted: Wed Nov 02, 2011 1:26 am

How did you come up with the 400 million records?

dick scherrer · Posted: Wed Nov 02, 2011 1:32 am

Hi Pedro,

Pedro · Posted: Wed Nov 02, 2011 1:41 am

Dick, sorry for my improper forum etiquette...

I was wondering how the original poster came to that number. It seems unusually high. I agree that a large number of files is un-managable. If there are really that many records, he will probably not even be able to browse the file.

I suspect is it is just a lack of experience in debugging.

Actually, it is not always straightforward to determine what kind of information to 'log' for problem determination. You probably need to re-create the problem several times. Each time you test, you narrow the area of testing and may introduce additional log records as needed.

dick scherrer · Posted: Wed Nov 02, 2011 2:31 am

Hi Pedro,

Bill Woodger · Posted: Wed Nov 02, 2011 4:34 am

I also find it difficult to believe that this file will "help" with debugging - just on the off-chance that you need it, as well!

I keep writing and deleting. It is just too absurd to capture how absurd it seems.

Has to be a better way. Let us know, please, let us know, what you are trying to do. Lots of detail. There has to be a better way. Can't believe you are even considering it for this purpose. Is you site made of money/DASD?

You could "easily" keep 1,000,000 lines in memory in Cobol, trap the abend, write out the latest million if needed. And that is a dumb idea.

gcicchet · Posted: Wed Nov 02, 2011 6:17 am

Hi,

if want to try something different, you can always use this