BethWilson
Profile Answers by BethWilson

Nov 5th, 2010

Interesting comparison. From my experience, and if those numbers are accurate, then there might be a?few other reasons:

1. You replaced the worst DataStage developers in the world with reasonable Informatica developers.
2. You replaced an application built years ago where the underlying architecture has changed, breaking a lot of the assumptions when the DataStage jobs were first built. Since DataStage upgrades fairly seamlessly, maybe nobody looked.?
3. I don't believe the custom coding or 6-to-1 mapping points. No one else says that, including the analysts.?This reinforces my?point #1.
4. 3-5 GB is not enough data to get performance differences in major tools.

On the general points:
1. Only the metadata repository has changed, not underlying engine. DataStage's metadata solution has expanded?way beyond ETL.
2. Any customer can use Oracle or SQL Server as the repository rather than the bundled, free DB2.
3. I once talked to a guy who complained bitterly about his DataStage performance in a table lookup of 120K rows. We added an index (no sequential scan) and the entire job ran 40x faster. Same would have been true for Informatica.
Had we put it into a DataStage DataSet or hash table, it might have run 100x faster.

It is confusing to choose, but good tools can be light-years ahead of hand-coding or bad tools. Good tools aren't that expensive from an ROI/maintenance perspective.
Take a look at this from your company's perspective rather than whether it widgets or wadgets and you will be better off.