Everybody has seen it: SAP cloud apps run on Oracle instead of SAP HANA, Oracle in-memory works instantaneously while with SAP HANA you have to rewrite the application and by the way Oracle runs twice as fast as SAP HANA.
This comes from a company whose CEO doesn’t remember SAP’s name and has said that this company hasn’t done anything in the last ten years. The ex-CEO claims that they are not monitoring SAP anymore as a competitor … interesting. A competitor whose name you forgot, a company, which hasn’t done anything in the recent past all of a sudden is worth front page ads in the Wall Street Journal? There must be more.
SAP HANA will soon pass 10,000 customers. Many large companies are migrating away from Oracle and putting their enterprise systems on SAP HANA. Do they want to advertise this? Definitely not, why should they? A database switch takes time and you need both suppliers as partners during this process. And here we are with SAP runs on Oracle…..The database replacement process is well under way, but not a day or an hour of services lost are acceptable. The analytical parts of the SAP cloud applications are running on SAP HANA. And the data entry ones will follow, no interruption of the services for the users allowed. Does SAP have to rewrite the apps? No, but Oracle doesn’t allow SAP to cross compile the hundreds of stored procedures and exactly these have to be rewritten. Other than that SAP HANA works nearly in full compatibility.
Can you write different and better algorithms with SAP HANA? Yes, because SAP HANA is so fast and needs only the columnar store in-memory and on disk in contrast to Oracle, where the primary data storage still is the row store and the columnar in-memory store works as an accelerator.
This means SAP HANA has a 6x smaller footprint on disk and 1.5x-2x smaller footprint in-memory and the traditional pre aggregation of transactional data can be completely avoided. This has a huge impact on the flexibility of the applications and is one of the primary reasons why customers are switching from ECC6.0 of the SAP Business Suite to SAP S/4HANA. As a byproduct backup/restore runs 5 times faster and systems for development and testing are much smaller. But perhaps Oracle likes the fact that their own systems or the ones which run on the Oracle database need more hardware to support the larger data footprint. Let’s not forget that Oracle bought Sun Microsystems a few years ago and therefore hardware is still one of its key businesses.
And Oracle wants to reverse exactly this newly acquired flexibility by dropping pre aggregation with the reintroduction of transactionally managed aggregates in the data warehouse. It is the wrong way to address the requirements of the future. Therefore SAP should let them march in that direction. The biggest challenge for enterprise systems in the future lies in their flexibility.
Reporting and analytics cannot be preconfigured and remain unchanged for longer periods of time anymore. To transfer the complexity from database structures to algorithms – using a variety of different hierarchies – is a breath of fresh air into enterprise computing, which today is still based on the management information system ideas of the sixties. Back then we believed all questions management will ever ask can be anticipated and therefore supported by hierarchical aggregations of transactional data, maintained in real time.
It is really flattering when the market leader in database technology pays you so much attention.
That shows how worried they are and how real the threat from SAP HANA already is. A new wave of changes in enterprise computing is happening right now and the early adopters will have a huge advantage in quality of products, customer interaction and service, optimization of processes and financial control. The simplification of the enterprise data models is one key element to become ready for the wave of changes. And while Oracle sees SAP HANA as a threat, our customers understand its potential and have already embarked on major application overhaul projects.
Oracle’s great concept that the application doesn’t have to change when the underlying technology changes, lasted for more than 25 years, but is coming to an end now.
Source: Blog from Hasso Plattner