Re: Import taking over 24 hours

Suggestions:

1) Try using the CPYFRMIMPF command using FLDDLM(*TAB) instead of writing
custom programs to convert from the tab delimited format.  Try this command
to see if will actually work with your data.  If it'll convert the data
correctly then check the performance of the command against these custom
programs your vendor has written.
2) If your target files have an index then modify the access path
maintenance on the target files from *IMMED to *REBLD or *DLY during the
load [CHGPF MAINT(*DLY)].  After you've loaded the data then change the file
back to *IMMED to cause an index rebuild to occur on the files.  If these
files have any logical files then just delete those logicals until after the
conversion or do the same access path maintenance modification on them.
3) Use blocking on the input and output files with the OVRDBF command using
the NBRRCDS and SEQONLY parameters on this command.
4) Make sure the conversion programs don't have some stupid coding like
declaring the input file as keyed when it's being processed sequentially.
5) General performance of the job.  Try bumping up the timeslice on the job
and not just the priority to give the job more system resources.  Maybe run
the job in a separate subsystem with memory allocated to that subsystem so
it gets some dedicated memory.

Paul

-- 
Paul Morgan
Senior Programmer Analyst - Retail
J. Jill Group
100 Birch Pond Drive, PO Box 2009
Tilton, NH 03276-2009
Phone: (603) 266-2117
Fax:   (603) 266-2333

> We had an outside vendor actually write the code for us which as I said
> before does work, but it is taking a very long time to process.
>
> If anyone has any ideas or thoughts on how we can speed this process up I
> would love to hear those ideas.  I need to check with the legal department
> before posting the code since I am not sure what contractual obligations
we
> are under as far as the code goes.
>
> I do know that he is splitting the fields up by checking for the tab and
> then writing the data to the file. the header file has quite a few fields
in
> it, and if anyone wants to know the number of fields I can verify that and
> re-post.