Hello, I'm looking for any advise in creating a model within Web focus using about 15 dimension sql tables and one fact table.

I've imported into web focus 16 master files. What i started to do is join the 15 dimension master files to the 1 fact table through the modeling view of Dev Studio.

What I'm starting to notice is when I create a fex file using this new synonym the performance is extremely slow. I'm assuming it's because of all the joins that it's doing.

The fact table and dimension tables are joined through the index field for each one.
The fact table has over 10 millions rows of data.

What would be the ideal thing to do here?
What I'm trying to create is a synonym where the user can navigate through the fact table using the dimension that correspond to it via a self service report.

Help is greatly appreciated.

This message has been edited. Last edited by: <Kathryn Henning>,

You need to build a datamart. The 15 dimensions (or some major subset of them) need to be combined into the main table. This may not be an option on your production database, but an overnight run that consolidates all of that material into a single table, a new table specifically designed to support your reporting requirements, will streamline everything.

What that does is take all of your execution woes and puts them on a task that runs at 2am when no one cares. When your users arrived at 8am they have a single table where the lookup values are already present in the core table. Add a few indices at the end of the nighttime run and you'll likely see a pretty brisk response time.

You can use WebFOCUS to do this, more likely Data Migrator, but I'd recommend you put the result into your big DBMS to get its industrial strength into the mix at reporting time. Your on-demand reports will run against a new master file that connects to the new deluxified table.

J.

Thanks John. So it appears that it would be ideal to have all of the structures done within my DB and have a synonym created on top of that.

From a tool perspective, when would you perform joins on synonyms? This seems that there are limitations to how many tables/rows can be joined?

I'm just trying to understand the methods availables, if I create a new table in our DB it would require that I maintain tables that would have duplicate data just because of the performance within webfocus?

if I create a new table in our DB it would require that I maintain tables that would have duplicate data just because of the performance within webfocus?

Yes. That's exactly what a DataMart is. They are not the "tables of record" but instead tables specifically designed to provide information via reporting applications quickly and efficiently. Their design often mimics that of the reports they support or in some cases actually ARE the reports they support. This isn't a WebFOCUS thing in particular. It's a common strategy across the industry.

Generally DataMart tables are updated via an overnight run that updates all new data from the previous day (or previous 30 days if you want to be more fault tolerant) or an overnight run that blows all existing DataMart records away and rebuilds the entire table from scratch. You can run it more frequently if you like. I have a Data Migrator process at a client site that runs every minute.

There's a couple of advantages to a DataMart. The first is that reports run faster. The second is that report requests don't tie up production resources that may need to respond to users in the field. With WebFOCUS there's a big third advantage -- you can design tables for InfoAssist that allow an end user to have everything they need in one table. One of my projects has things like County Names and Vehicle Makes all coded instead of written out. My reporting tables don't show a County of "22" or a Car Make of "14", they show "Washington County" and "Ford" which is infinitely more understandable to an end-user.

J.


By the way -- joins in synonyms -- I don't do 'em. If I can build a view (in the RDBMS) to cover the join or create a reporting table for heavily-used connections I'll do it. If I need a join for a specific focexec I'll do it on the fly in the source code. At that point my master file is flat, which makes for easier code maintenance downstream in the lifecycle. Joins in master files tend to be overlooked when something goes wrong in a report output.

oh ok. I am using a data warehouse in which I bring in all of the data we have collected from the different systems we need data from. Our warehouse is generated nightly with new/updated data.

We are building fex files on top of the data warehouse, so we are never hitting the production system of use.

We have just recently acquired web focus, so I'm going through the setup of these synonyms and trying different things to create a self service reporting solution for the business. It does seem that the structures need to be created in the RDBMS instead of doing the joins within web focus.

thanks John, i'll look into this and let you know what we decided to do and how it went.
thanks.