Sep 27 2011
HANA Like An iPod? More Like A Digital Camera…
Timo Elliott published a great blog post this morning:
Why In-Memory Analytics is Like Digital Photography: An Industry Transformation
Timo is an avid photographer as well as a BI evangelist, and in this post he combines his knowledge of both, making some excellent points along the way. It’s well worth hopping over to his blog to check it out.
Its a nice article. I like the innocuous bit ‘Once you have the row-level data in-memory’. I am a little worried that organisations will think that technologies like HANA are an excuse not to have a data warehouse.
According to some folks at SAP, that’s exactly where they’re going. No data warehouse. 🙂 Today we need warehouses (and all that comes with them, such as scheduled ETL, data validation, and so on) because we don’t want to cause issues on our production systems when we run analytical processes. With HANA it won’t matter. You’ll run your analytics on the same hardware (same database) as your application systems.
We’re not there yet, but that’s one of the components of their long term vision.
They are putting a lot of faith in the structure, quality and completeness of the data. I still see a place for the disciplines of data warehousing for bringing together the data in a way that makes sense to end-users and can be trusted as the truth. Technologies like hana and, to a degree, qlikview are competing with OLAP to provide data faster than a regular RDBMS, but the data will still need to be collected together and enriched before (or perhaps, while) you let the in-memory engine at it.
My thinking is that the data warehouse will eventually become a repository of data enrichments (calendars, dimension groupings, etc) that dont live in the source systems that is then joined in-memory to the source data to provide a user-friendly reporting source.