HANA on Power? It May Be in the Cards

January 15, 2013 Alex Woodie

SAP made a major splash in the IT world last week when it announced that its HANA in-memory database can now be used to power not just analytical workloads, but transactional workloads in its ERP suite, too. It’s a notable technical achievement for the German software giant, but it left SAP’s IBM i customer base with little to applaud, since HANA currently runs only on X86 servers. But that could change if the work that SAP and IBM are doing to get HANA running on Power pans out.

During last week’s webcast from Palo Alto, California, SAP technology and innovation executive Vishal Sikka told his audience that SAP and IBM are currently working on getting HANA running on the Power processor. “We’re also evaluating the work that we’ve been doing on Power to see how far we can go with Power, the work that we’ve been doing jointly at HPI,” Sikka said. (HPI stands for the Hasso Plattner Institute for Software Systems Engineering.)

Ken Tsai, vice president of SAP Hana product marketing, also confirmed the news in an interview with IT Jungle. “Power is something that we’re looking at very closely right now,” Tsai said last week.

However, SAP is not releasing many details of the work it’s doing with IBM. When asked elaborate on the possibility of getting HANA to run on Power and what kind of work that might entail, SAP responded with this comment from Amit Sinha, head of database and technology product marketing. “[HANA] on Power is a research project currently sponsored at Hasso Plattner Institute. We await results from that to take next meaningful steps jointly with IBM,” Sinha said via email.

There are several reasons why the combination of HANA and Power would be tempting to SAP and IBM. The most compelling may be the simplification story. Both vendors are spending a lot of time and money these days to try to simplify their customers’ IT environments–SAP by putting HANA in charge of transactions and analytics, and IBM through its PureSystems initiative to create “expert integrated” systems.

According to Sikka, HANA simplifies a database administrator’s job because “a lot of tuning … techniques we have been used to in traditional databases are not needed in HANA, certainly not to the same degree as in traditional databases,” he says. “In fact we’ve seen administrators of traditional databases easily not only becoming administrators of HANA environments, but able to do innovation in HANA environments.”

If HANA is successful at driving a lot of complexity out of database administration, then it will be a perfect fit for PureSystems. In fact, IBM would probably want it running on as many of systems as possible, including the Power Systems servers as well.

SAP Business Suite currently runs and is fully supported on IBM i and DB2 for i (DB2/400). There are most likely somewhere between 1,400 and 1,800 or so organizations running SAP on IBM i, which makes SAP one of the platform’s most popular, and modern, ERP systems. While exotic databases can be found powering some esoteric ERP packages, SAP is the only big ERP vendor to take the plunge and make the case for ripping a relational database management system out of ERP. SAP had been telegraphing its intentions to do this for year, but it’s still a gutsy call to actually do it, and it might just herald the start of a new era for enterprise computing.

It probably doesn’t surprise the astute IT Jungle reader that SAP on IBM i environments are already smaller, more streamlined, and easier to manage than SAP on Wintel environments. In an interview with Dan Burger last year, Ron Schmerbauch, technical leader of IBM’s SAP on i team, attributes this to the nice meshing of IBM i subsystems to SAP.

If SAP on IBM i is already quite streamlined and helping keep customer costs down, why wouldn’t IBM and SAP want to take simplification to the next level and help customers even more? Answer: they most certainly would opt for more simplification.

Getting HANA running on Power would also enable a joint nose-thumbing in the general direction of Larry Ellison and Oracle. If IBM is already losing SAP database deals to HANA (a somewhat arguable fact given that DB2 for i is sold as part of the IBM i OS, but let’s not dive into accounting codes), then at the very least it would want to sell some hardware to recoup its lost database revenue. Oracle already stands to lose a lot of money if HANA takes off as the transactional heart of SAP installations. With HANA running on Power–that is, with HANA running on PureSystems–IBM and SAP could offer a very complete business solution.

The question remains whether SAP and IBM will spend the money and the time to get HANA running on Power. It would probably have to be a separate version of HANA, since even IBM has different versions of DB2 for its X86, Power, and mainframe platforms. HANA was written in C and C++, so it’s technically feasible, provided they invest in engineering HANA to run on the Power architecture.

At the very least, getting an in-memory database on Power that can drive not only analytics but transactional workloads, too, would be compelling enough to grab IBM’s attention, which it apparently already has. Stay tuned for more news as we learn it.