The King is dead. Long live the King. SAP HANA has arrived (actually that was a good 12 months ago) and served notice to disk-based relational databases.
With the forthcoming release of SPS03, BW running natively on the HANA in-memory database goes into GA. And as early as Q4 this year we could see ECC running on SAP HANA, and thus a single database engine writing simultaneously to row and column stores for transactional and analytical usages. The nirvana being of making informed and rapid decisions at the speed of thought is at hand.
Oracle has subsequently responded with the incorporation of the TimesTen In-Memory Database within its Exalytics appliance. The merits of one over the other has been the subject of some lively polemics.
SAP HANA does away with the notion of extensively transformed physical data models – star schemas – to serve specific reporting requirements. No longer does the data warehouse necessitate a layered physical data model to provide the conformed view for business analytics. In the traditional spinning disk world, overnight batch jobs chug away in transforming, consolidating and aggregating data. With read-optimised columnar tables materialised in-memory, all the processing can be done at query runtime at blisteringly fast speeds.
Ralph Kimball’s dimensional star schema and Bill Inmon’s notion of the enterprise data warehouse and corporate information factory will continue to be relevant; but in more abstracted forms where the data is persisted once and the business logic modelled within the OLAP layer.
SAP has an established track record of doing in-memory with APO Live Cache, TREX and BWA. In fact much of SAP HANA’s columnar heritage also comes from the TREX engine. The Sybase acquisition beefed up the SAP’s columnar expertise with the best of breed disk-based Sybase IQ database coming into the SAP database stable along with some very smart engineers. Moore’s law has brought the affordability of multi-core processors to a feasible price point. An axiom that will only continue to drive the adoption through the enterprise and SME space.
So how to make sense of SAP HANA? How does it add value to the bottom line? SAP HANA is licensed by RAM usage in 64Gb increments and even the most minimally sized deployment represents a signficant CAPEX. And that’s before we consider the most appropriate middleware and BI presentational tools.
For the answers we need to look in the right places and that’s not at the technology but the business. That’s to say find the unequivocal business case. Ultimately SAP HANA is just a database running on highly tuned bespoke kit. What business pains does it solve? What competitive advantage does it enable?
These are questions that only the business can answer – albeit with close collaboration and facilitation from IT.
As always thoughts and opinions welcome!