Hello,
We are upgrading and have the opportunity to improve on our process and report performance. What is the advantage of joining data sources to creating master files verses multiple hold files that pull one at a time within a single report. Basically create the entire fex as a master file verses a hold file. Both would have the same criteria as a parameterized report.

Yes, some of the joins are used in other reports but the reason for multiple reports is the uniqueness of the data provided.

The Question: Which will would be the best method to use, make all the joins as master files or join in each fex?

Thank you,
Geri Gellman

This message has been edited. Last edited by: Tamra,

What's your DATA SOURCE? Could your specific requirements be translated into an RDBMS view?

I don't think a join in a master is any more or less efficient than a join in a fex. The benefit of having it in a master file is that it doesn't need to be repeated in the fex and that promotes consistency and makes it easier to debug and update.

The Data Source is typically all SAP which has many, many tables that we must join together to get anything usable.

Not familiar with RDBMS view. Is there a manual on that option?

Thank you,
Geri Gellman

An RDBMS View is basically a SELECT with all the joins already coded. To WebFOCUS it'll just look like a table, but all the joins are already set up in the database and optimized. It's a way of creating an abstraction for end users to access the data without having to remember or even know what fields to use and what type of join they need.

However, if your data source is SAP, then the chances of your being allowed to create such joins is very low, which explains why you have all these hold files.

Anyways, like I mentioned, I don't think you'll see any difference in efficiency of joins in a master versus a fex. However, it's always best to have one place for all these joins for maintenance purposes which means you should in fact put them in your master file.

Thanks, I just wanted to check if anyone thought one process might be better than another!

I appreciate the help.

quote:

Thanks, I just wanted to check if anyone thought one process might be better than another!I appreciate the help

Hi Geri,
below are the few steps which I follow in performance.
1. Use Cluster Join on master file. Avoid Report level join. Cluster Join creates Run time join which helps to increase report performance.
2. if you are hitting DB server multiple time for calculative date. Try to hit single time and get data into hold files and perform operation on hold file.
3. Use proper Group By, Order By and conditions. so that you can retrieve data faster.
4. And if possible use Dimensional Modelling(fact and dimensional tables ).