Hello,

I have about 150 teradata BTEQ scripts run on a nightly basis,
some to do updates/inserts/etc from other various tables in order to
populate/maintain/update our data warehouse. (If your unfamiliar with
a td bteq script, i think its similar to a DB2 CLP script)

Being typical TD BTEQ scripts, each one has information such as
"login/password" on the first line, some simple error trapping and
notification messages in each script, and quite a bit of code copied
into each of the scripts.

In retrospect, after development, I wonder if there are better
solutions such as a single application to dynamically create text from
a single script/application and execute the script. This way,
isolating much of the code/testing in a single place and abstracting
the business logic for DML behaviors for the data warehouse.

Does anyone have any experience with maintaining many ETL scripts and
what levels of abstraction have you including in the scripting?

Any suggestions appreciated.

TagSoup (AT) gmail (DOT) com wrote:
Coming from an Oracle perspective, a lot of common functionality was
kept in stored procedures in big projects I worked on,especially the
Translation step logic. Source control outside the DB, and Oracle has
PACKAGES which group functions together, so every package included a
WHOAMI function that returned revision information.

HTH,
Ed