SAP ups ante on in-memory computing appliances

SAP strengthened its eager embrace of in-memory computing on Wednesday, announcing plans for a series of applications that include programs it calls previously "unimaginable" as well as some that overhaul existing software.

In-memory computing moves data off traditional storage and into RAM, providing a performance boost over reading data off disks. SAP's in-memory data processing engine is the brainchild of co-founder Hasso Plattner and at the core of the HANA (High-Performance Analytic Appliance) that SAP released last year.

"An entirely new way to construct applications is in front of us, and it is without compromise," said SAP executive board member Vishal Sikka during an event in Boston on Wednesday.

New in-memory applications planned for release this year include a tool for helping operations and marketing departments work together more efficiently, SAP said.

Another new offering will help finance workers crunch data from open invoices and determine, using both current and historical data, the type of discount terms a given customer should receive. A third planned release will provide analytics for smart meters, SAP said.

SAP is also planning to revamp its Trade Promotion Management and Cash and Liquidity Management applications using in-memory technology.

Specialist analytics

Overall, the announcements are a continuation of SAP's previously stated plans to release specialised analytic software meant to sit alongside a customer's existing ERP (enterprise resource planning) installation, pulling in data in real-time. The company announced an initial release, Strategic Workforce Planning, late last year.

Although SAP has repeatedly touted HANA's high performance, it has other goals in mind as well, such as making it easier to deliver analytics to end users. SAP demonstrated a number of upcoming tools, including a graphical report builder that one executive dubbed "a blank canvas for analytics."

Using it, a worker could make natural language queries for business data, such as "sales per country," which would then be served up from a back-end information store. Then, the user could make simple drawings that would generate various visualisations of that data. For example, drawing a circle would generate a pie chart.

In-memory computing also stands to cut down on complexity in IT environments, Sikka said. He described a customer that is running SAP's Business Warehouse, Oracle databases, specialised hardware and SAP Business Objects software for BI. SAP's in-memory approach could reduce the number of required layers, he said.

The in-memory applications SAP is building amount to "a new cartridge inside of HANA," Sikka said.

Still, so far SAP is not positioning its in-memory engine as a full replacement for customers' main database, which is Oracle in many SAP shops. But that may change as SAP's technology matures.

In the meantime, SAP's in-memory application strategy stands to give it products to sell that customers seem eager to buy, as business intelligence sales remained solid during the global recession.

Work still to be done

While SAP has laid out an ambitious vision for in-memory technology, and has a broad foundation on which to build next-generation capabilities, much work lies remains to be done, according to Forrester Research analyst James Kobielus.

For one, SAP has two sets of data-processing technologies, with HANA, Business Warehouse and Business Warehouse Acceleration on one hand and the Sybase IQ database on the other. The products currently have minimal integration, although future releases will address that issue, Kobielus said.

A step in this direction will occur later this year when the second version of HANA is released and SAP's Business Warehouse platform is ported to HANA, simplifying its deployment, according to Sikka.

HANA in its current incarnation is also limited by its primary focus on existing SAP customers, and technical limitations such as a lack of "query predicate pushdown and efficient compression to an intelligent storage layer (which are core features of both Oracle Exadata and IBM Netezza TwinFin)," Kobielus said.

In addition, HANA so far "does not address the price-competitiveness issue that prevent SAP from competing head-on with Oracle Exadata and IBM Netezza TwinFin for new accounts," Kobielus added. "SAP has not addressed the need to provide low-cost [data warehousing] appliances for the midmarket."

It's also likely that SAP is not done building out its family of analytics-related technology, and will make acquisitions in areas like information lifecycle management and even hardware, Kobielus said.