Recently in Data Warehousing Category

Having now had this problem installing the Oracle Warehouse Builder with version 11g and 10g I though it deserved a mention here in my blog. If the Repository Assistant is just vanishing around 87 / 88% percent and you can see a final entry of "I am in processLoadJavaToken" or under unix you may get a message from the assistent 'reposinst.sh' saying "Failures occurred during processing" this might be helpful to you.

What is going on is, that part of your xml code is corrupt in the database, but that is not apparent! All you need to do is go to your installation directory "ORACLE_HOME" and switch to rdbms/admin then de-install and re-install the xml code like this:

1. log in as sysdba ("export {SID}; sqlplus / as sysdba")
2. enter "@$ORACLE_HOME/rdbms/admin/rmxml" and press enter. Wait for it to finish.
3. enter "@$ORACLE_HOME/rdbms/admin/initxml" and press enter. Wait for it to finish.
4. exit

After that remove any repositories before re-running the assistant. Also remove any roles installed by OWB.

Well in case we were not already convinced, a report out this week put Oracle in the driving seat with regards to market share for Data Warehousing Tools.
Oracle apparently has 40% of the market which was around $1.8 billion in 2006. This is evidently a major area for investment focus within Oracle as is indicated by the Oracle Business Intelligence Blog.

OK we all know there is no such thing as success on demand, but given the right set of tools and the right set of rules it should be a darn sight easier. This article by DMReview gives you some real show stoppers to avoid. If you can adopt these rules or at least adapt them to suite your business you'll certainly be on the right road.

This article gives a rather neat insight into what you should be looking for within your Data Warehouse in order to meet the Basle II requirments. If you're not sure if you're on track take closer look.

This article in the DMReview underlines one of the realities we have experienced in the recent past. Its usually faster and more effective to load first and then transform the data within the RDBMS. This gives you much greater potential with regards to analysis and transformation.

This article by DMReview gives you a simple but affective list of 10 mistakes which you shouldn't make when developing a Data Warehouse.

1. Leaving out the users


2. Testing reports - not data


3. Skipping the reconciliation of data warehouse data with source system data


4. Underestimating the testing workload


5. Failing to set proper expectations with business users


6. "Fast tracking" the testing phase


7. Underestimating the duration of the ETL process


8. Choosing the "wrong" testers


9. Skipping proper sign-offs


10. Using "made up" test data