Mission critical HANA 2: SAP bulks up on high availability features
The next release of SAP's in-memory HANA database plugs a few gaps for enterprise high availability features. As the in-memory database market commoditizes, SAP differentiates with analytic capabilities for extended data types.
While SAPHANA helped pioneer the modern in-memory database back in 2010, in years since the declining cost of in-memory based systems has no longer made HANA such a novelty. For its next act, SAP steps up its high availability, storage tiering, database architecture design, plus additions to the analytics function library, among others.
At SAP TechEd Barcelona this week, SAP is announcing the broad outlines of the HANA 2 release. What goes into the final package won't be known until it goes GA on November 30 - SAP is looking to see what features it can get into this package.
But SAP is planting its stake with enhancing high-availability by adding a new active/active mode that allows read-intensive workloads to be diverted to a secondary server without taking the system down. It's a feature expected of databases supporting heartbeat applications; by comparison, Oracle requires buying an enhancement of its high-availability tool.
SAP is also streamlining the capability to tier data between memory and disk; although the details have not been announced for this release, we expect that it will probably have some form of rules-based capability for automating the shift of data that gets shunted off memory to disk.
Other highlights include a new version of SAP's Architecture Designer modeling tool that allows you to reverse engineer table layouts from other databases such as Oracle or Teradata onto HANA. This tool complements the existing HANA Studio tool that defines and lays out tables in HANA.
For the new release, SAP is bulking up the predictive analytics libraries supported in HANA with eight new ones targeting classification, association, time series, regression analysis, and summarized statistical functions. In turn, SAP has added some analytic microservices in the cloud for performing text analysis, entity abstraction, fact extraction, and linguistic analysis,
There are several goodies for developers in HANA 2. There is an extended application server that provides developers to work with their programming language of choice, while the HANA Web IDE simplifies modeling spatial, graph, and text data entities.
Database migrations can be disruptive, especially if you have heartbeat applications like ERP or SAP BW running on them. Acknowledging that, SAP will continue supporting the last two versions (HANA 1, SPS 11 and 12) for those portions of the client base. But migration to HANA 2 (which will be free) will require HANA 1 customers to be on the last (SPS 12) version.
SAP originally differentiated HANA with the ability to perform transaction processing and analytics on the same engine. To some extent, established players including Oracle, IBM DB2, and Teradata have also grafted in-memory analytic column stores to their flagship database platforms. HANA still differentiates by being a pure in-memory play; but then again, rivals such as MemSQL have also added analytics to their transactional repertoire. To that, SAP emphasizes its capabilities to hold extended data types, such as graph and time series data.
What's significant is that in the current release, SAP's prime strategy is bulking up the enterprise high availability capabilities as it increasingly positions HANA for mission-critical workloads.
But keep your eye on SAP's progress in bringing recently acquired Altiscale into the family. With its dedicated Hadoop cloud hosting infrastructure, the Altiscale deal could help SAP bolster blend HANA with a Hadoop implementation that it owns, which could significantly scale HANA'S analytics value proposition.