Hello folks,
i have two hold files each having around 10-15 million records. when i run these hold files separately in deffered status its taking me around 30-40 mins to get the result. now the problem is i am joining these two hold files based on equijoin condition and when i run it as deffered i am not able to view the result even after couple of hours. any suggestions so that i can improve the run time of the report after joining the hold files and also when i try to view the deffered report(after 2-3hrs of runtime) its taking lot of time to show the results and i am getting session time out error.

thanks
maddy

What is the format of the HOLD files?

You should at the very least, create FOCUS database HOLD files - your JOIN will be very much more efficient than flat files. Create indexes and key fields for the JOIN fields.

hi francis,
Thanks for the suggestion.my hold files are in focus format and also i created indexes on the join fields and still i am facing this problem.
and also when i try to view the deffered report its taking lot of time and i am getting session time out error.
thanks
maddy

I think you need to summarize your data at a higher level to create less rows - 10 million rows in a join is not very efficient. Perhaps you can create a drilldown type report - start at a very high level and drill down the lower levels...

We had a situation in our conversion from mainframe to WebFOCUS in a program running forever and ever as in four or more hours in deferred mode. Turned out we were not executing the version we thought we were. I don't know as this would be your problem, but you might ensure what you are executing is what you want to be executing.

Our data is in DB2, conversion has proved interesting over the past couple years and we haven't gotten to any of the fancy stuff yet.

Maddy,

You're joining two FOCUS files with 10-15 million records each? Yowee. It's not too surprising that it would take a couple of years for that.

What kind of join do you have - one to one or one to many? If it happens to be one to one, you might actually do better creating flat files and then merging them using MATCH FILE. Not that I'm a fan of MATCH FILE, mind you. If it's one to many, you might get a better result by creating and then joining two flat files, instead of two FOCUS files (make sure they are sorted in the same order). I know that sounds ridiculous, but sequential processing of two ordered files might work better than indexed lookups in this case. Might not. Can't hurt to try it, though. Good luck.