Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations Mike Lewis on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Detail hidden in Transformer server build 1

Status
Not open for further replies.

doddy

Technical User
Mar 28, 2001
19
0
0
AU

I have a cube that builds perfectly on the desktop, but fails on our UNIX server - actually it still builds, but has not "connected" the data correctly.

A group of star-schema tables are used. Each dimension has a surrogate_key at its lowest level, and the fact table query for measures data includes all of the foreign surrogate_keys.

The server build opens in Upfront looking fine (all totals OK) but if you attempt to split the totals by the dimensions (eg. split total products by their product type) the detail area comes up all zeroes. It's as if the server build has not matched the keys correctly.

All queries deliver, all keys should find matches.

Access Manager security is built-in, but I have top-level access so should be able to see all data splits. We're using Transformer 7.0.

Your help would be a wonderful thing.

regards,
Doddy
 
Check the output scale on your measures. Sometimes the scale will be used correctly on the desktop, but when you build in unix, the scale applies incorrectly, or vice versa.

We've run into this before, usually its a matter of either Unix or Desktop incorrectly interpreting the datatype. Multiplying your measure by 1 in your IQD usually solves this misinterpretation.

Pain is stress leaving the body.

DoubleD [bigcheeks]
 
DoubleD, you champion. A star for you.

Sadly, I did the unscientific thing and adopted both of your suggestions at once. Consequently, I don't know whether it was multiplying the measures (and numeric surrogate_keys) by 1 in the IQDs, or setting both the input and output scales to 1 to ensure the numeric type was brought right through to the measures, but everything has come up a treat.

Now about this world peace ...

regards,
Doddy.
Adelaide, Australia
 
Depending on what you are trying to accomplish, using both suggestions is valid. They really deal with two separate issues. The multiplying by 1 forces Transformer to recognize the field as the proper numeric datatype. The Input & Output scales allow you to decide how you want your data displayed in your cube.

Glad it worked for you, I'm still working on the world peace issue, I'll get back to you when I have a clearer vision.

Pain is stress leaving the body.

DoubleD [bigcheeks]
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top