Can SAP HANA boost Real-time Data Analytics?
In a recent press news from 13th December, SAP announced at the SAP Influencer Summit in Boston that “leading software vendors are adopting the open SAP HANA platform for their existing products and building completely new applications.” Among them are companies such as T-Mobile and TIBCO. The In-Memory-Technology HANA has been introduced in 2010. SAP says, the DBMS technology is at the core of its platform roadmap and supposed to support their cloud and mobility strategy.
The HANA platform is used within real-time analytics applications. For performance optimization, the platform reads large amounts of operational and transactional data from any data source into memory – all of that happening in real time. One class of applications that can leverage such capabilities are Smart Grid applications. For example, by analyzing huge amounts of smart meter data, utilities could reduce costs by yielding more precise consumption forecasts. SAP claims in its advertisements that depending on the system configuration data analyzes can be up to 3600 times faster.
It is important to consider that HANA is work in progress. Its first ingredients comprise the Business ByDesign Software Development Kit, the River application Platform as a Service offerings which is language-independent but Java-based, and an analytical appliance.
According to a technology report by Gartner Research,
The HANA Architecture is a work in progress, and will undergo several significant changes before it is completed. This potentially exposes SAP users to challenges for migrating to, and integrating with, different technology generations.
SAP's vision will force megavendor competitors to respond by clearly and openly articulating their cloud and in-memory computing strategies.
In addition to addressing formidable technology and go-to-market challenges, SAP will have to introduce its highly innovative vision in a way that will be acceptable to the most-conservative part of its customer base.
It needs to be seen whether the strategy to hold large amount of data in memory will really work, as some critics doubt. Eventually, the press release tries to address these doubts by introducing success stories. For example, Jeff Wiggin, vice-president of Enterprise Information Technology, T-Mobile, USA is cited:
We recognized that being able to respond to the needs of our customers in real time would give us an incredible competitive advantage and improve the quality of the customer experience. In order to deliver that kind of experience, we needed an underlying platform behind our sales and marketing efforts, allowing us to uncover customer insights and then act on those insights in minutes, not weeks. SAP HANA delivered exactly that.
T-Mobile is using HANA for storing 2 billion customer records to run all necessary reports with an average response time of five seconds.
It is surprising to see the speed at which even more conservative IT vendors are now fostering their cloud strategies.
Readers interested in details of the technology topics behind HANA can read the book “In-Memory Data Management” by SAP founder Hasso Plattner and Alexander Zeier. Moreover, Gartner published a technology report on HANA two months ago.
No need to pre-compute OLAP Cubes
In the traditinal model, summarized views (OLAP cubes) are batch built periodically and then used for analysis. Not only is there a delay in getting the latest usable data (a few hours late just does not cut it anymore!) but also you are stuck with existing views of the summarization which can't be changed easily without DBA involvement and such.
HANA offers some of Hadoop's capabilities but processing is much faster in the case of HANA. A Hadoop cluster can ultimately be scaled beyone what HANA is built for. The largest Hadoop cluters span 1000's of machines and are into the multi-petabyte range.