Thursday, September 1, 2011

Tools in the ETL Warehouse

Getting an ETL team to start using a new product is usually a tough task. Employees comfortable with scripting usually will find any excuse to throw in a 3 page SQL override or call a Unix shell script from the command task. This totally negates the use of the ETL Tool (ex: Informatica).

In an effort to bring conformity, and to promote developers to use Informatica, some clients make a blanket statement to use only the ETL tool for "Everything".

It may sound all uniform and nice to hear, but I think this is'nt the right approach. The ETL tool (Informatica) is a data migration and integration tool. If you ask me, its like asking the carpenter to use only a Hammer for all his tasks. It will not work! The processes and the number of objects needed to accomplish even the simplest of tasks will be unnecessarily complicated.

For example, the components needed to put together a mapping to generate a parameter file, could be replaced by a ten line shell script.

In my opinion, a good standard on use of ETL tools is to mandate that Any data movement will only be done through the ETL Tool. The other components that enable the ETL tool will have no part in data manipulation or extraction.

If you ask me that is...

No comments:

Post a Comment