IBM i Shops Not In A Hurry For HANA
May 11, 2015 Dan Burger
The IBM i community has always looked to IBM for evidence that it is protecting its collective investment in the platform. Lately it seems the IBM i executives are playing up the investment angle more than usual. At the COMMON Annual Meeting and Exposition, Dave Nelson, director of IBM i development, and I talked about that topic. Nelson played a key role in the recent Power-SAP Summit in Rochester. The event was primarily for IBM i business partners and SAP end users and their interest was in roadmaps.
There’s an existing perception among the SAP on IBM i community that SAP wants to force users to adopt HANA, the in-memory database that last week became officially supported on Power Systems. Although SAP markets HANA for business intelligence workloads, it has also repeatedly stated that its goal is to eventually replace all underlying relational databases.
And there aren’t any specific plans to run HANA on IBM i at this point.
There are IBM i shops interested in HANA, but most plan to stay the course they are on as long as IBM can guarantee feature and performance enhancements to DB2 for i will keep that database competitive.
SAP has poured a great deal of effort and money into creating and marketing HANA, but a single-minded strategy would not be wise at this point. There will continue to be a majority of SAP customers that are not ready to implement HANA, but that still require attention. And, of course, the revenue stream from this set of customers is significant.
Platform loyalty is a strong bond in the IBM i community. It can be a blessing and a curse, but it’s not as if the SAP on IBM i organizations are stuck in the past. It’s not difficult to find projects that involve new technologies such as independent auxiliary storage pools, solid state drives, PowerHA clustering, PowerVM virtualization, and live partition mobility.
“We understand what the HANA marketing is, but we know when we run those same types of workloads we see things to be fairly comparable,” Nelson explained. “If they have SAP on i today, it doesn’t make sense just to go to HANA because SAP is telling them to go to HANA. In the future . . . maybe, but today they are competitive and there is a lot of work to go that route.
“There’s not a dire need for our IBM i customers to move to HANA anytime soon,” Nelson continued. “We, the IBM i development team, know the core services from SAP have to have the requisite function and performance to remain competitive. We’ve run those tests and we are competitive. But anytime you run a handful of queries, you will be better on some and have room for improvement on others. That’s true for any database and any set of queries.”
As for IBM’s commitment to investing in IBM i, Nelson points to the stream of database enhancements that have raised the bar on both function and performance with each Technology Refresh announcement and OS release. No one is saying how often DB2 database enhancements are suggested by SAP or the SAP on i customer base, but can you call it a coincidence that most of the enhancements don’t require SAP to make changes to its applications to leverage the new stuff? Credit some of this to the synchronization of product roadmaps between IBM and SAP.
“We have always been very competitive with OLTP performance and we continue to make enhancements with the OLAP and Business Warehouse (BW) performance in DB2,” Nelson said. “We’ve had performance gains in OLAP queries, technologies like encoder vector indexes have been in place, and SAP leverages this.”
Those organizations that are leaning in the direction of HANA believe there are performance advantages compared to an OLAP business warehouse environment.
“I have not seen any statistics that show HANA versus SAP on i is an advantage,” Nelson says. “Some customers were asking for that during the Summit, but there is nothing available. We have our own benchmarks that focus on Business Warehouse tasks, but you will always find that mileage varies from one customer to the next.”
IBM clearly wants its IBM i customers that run SAP to remain on Power Systems. But if and when IBM i shops decide they need to go to HANA, IBM wants to make sure they can remain on Power.
For now, Power-based HANA runs on SUSE Linux and handles data warehousing, analytics-only workloads. There are two Power Systems boxes configured to run HANA. The smaller system is a Power S824 optimized to run HANA workloads exclusively. The larger system supports HANA in PowerVM virtual machines and allows for additional workloads to be run on a Power E870. The servers are designed to store archived data from that which is hosted in memory data on SAN arrays rather than internal disk.
If HANA eventually supports IBM i, and there’s no indication that will happen, the level of complexity seems guaranteed to push this endeavor well into the future. Expectations can change, but the roadmap seems to lead to a Linux-only conclusion for HANA on Power.
There’s also the consolidation motivation, which is a factor IBM is very much aware of as it promotes SAP on Power. Consolidation has been an important message IBM’s delivered to the IBM i community for years. For shops that have SAP running on more than one Power platform (IBM i and AIX), the choice to add Linux-based HANA as part of a Power consolidation move, fits nicely with the IBM vision.