31. Januar 2010
Interessant finden wir Tom Kytes Meinung zu dem Thema ETL-Tools, hier stellvertretend für PowerCenter Informatica:
Management is mis-speaking. It is not that “business and data transformation rules” would be ‘buried’ in plsql – it is that since they are using informatica, things written in plsql would be “outside the realm of the meta-data informatica keeps”
To me – the primary reason to use an ETL tool is “documentation and maintanence”. You will likely find the tool getting in the way in many cases (times you just want to through your hands up and say ‘lets just code it already’). But the tool isn’t there necessarily to make this aspect easier – it is there to provide continuity, documentation, change management, the ability to figure out where the data comes from and where it goes. This is especially important for future generations of developers that will follow you (and want – no, need – that meta data).
Probably not what you expected me to answer
We have our own ETL tool – we call it OWB (Oracle Warehouse Builder). It happens to generate plsql and oracle specific SQL (and does things in SETS which is great). But it has the same sort of meta data approach. We could argue which is better – Informatica vs OWS but at the end of the day – you are
using them not necessarily because it’ll make the entire ETL process easier or even faster – but because of the documentation they provide. On a large scale effort – with many developers, lots of data sources and a lifespan that should be “long” (eg: maintaining this system is relevant!) – they make sense.Zitat von Tom Kyte, ORACLE