How SAP HANA Helps Keep IBM i Strong
February 24, 2020 Timothy Prickett Morgan
It has been a good 12 months for Power Systems in the cloud. Not only did Big Blue launch entry Power8 and high-end Power8 machines on its IBM Cloud supporting IBM i and AIX and promise to get high-end Power9 iron on its cloud as well, but Google, Skytap, and Microsoft also launched Power9 iron on their respective public cloud and offered to run Linux, AIX, and IBM i on these machines. Last week, it was the turn for enterprise software giant SAP, which is adding high-end Power E980 systems to its own cloud services so customers can run HANA on Power machinery instead of X86 iron.
Neither IBM nor SAP talk much about how much their respective businesses drive each other, but every now and then we get an inkling here and there. We have been particularly interested in how HANA, the in-memory database that SAP created as its future platform, making something akin to a real “Application System” that runs atop Linux and that includes HANA as well as the S/4HANA application stack from SAP and any homegrown applications that customers create themselves to work on HANA. The underlying operating system is Linux, of course, but other SAP applications can run natively on IBM i as well as various Unixes and Linux, so it is not an either-or proposition with SAP Business Suite. Many times, customers can run SAP applications and HANA side-by-side in logical partitions or across Enterprise Pools of multiple Power Systems machines.
As you know, we are always interested in anything that directly helps bolster the combination of IBM i and Power Systems, but we are equally interested in anything that helps Power Systems maintain or expand its share of wallets among the datacenters and dataclosets of the world. As we are fond of saying: Anything that makes Power Systems stronger makes IBM i last longer, and like all of you reading this publication, we here at IT Jungle certainly have a stake in this business being a healthy one.
Over the past decade and a half, IBM has added thousands of SAP Business Suite customers to the IBM i rolls, and some of these customers are among the largest IBM i shops. SAP has over 45,000 customers running its various ERP software worldwide, as we recently reported, and about 11,500 of them have moved to the S/4HANA application and HANA database combination. This is a slower transition than perhaps SAP had hoped, since it wants to control the entire stack – database and applications – at its customers and not watch as Oracle, Microsoft, and IBM make all of that money on databases. The core operating system isn’t worth all that much, at least not compared to the amount of dough that is spent on the databases and applications.
Maybe IBM should have bought SAP? But anyway. . . . Still, having about a quarter of the customers move to S/4HANA and HANA is not too bad. The other three quarters could take half of a lifetime, or a lot less time.
When HANA first rolled out in stages between 2010 and 2013, it was only available as a service on the cloud. When it was available on premises, it was available only on X86 servers, and very specific ones that had been vetted by SAP itself with very precise configurations of compute and memory and with only SUSE Linux Enterprise Server to limit the support matrix and therefore increase the odds that HANA would behave properly at IT shops that were out there on the bleeding edge. In May 2015, when IBM worked out a deal with SAP to allow HANA to run on Power Systems, Big Blue used a packaging and testing regime called Tailored Datacenter Integration, or TDI, that allowed HANA to run on logical partitions on Power8 servers instead of bare metal iron. According to Vicente Moranta, vice president of offering management for Enterprise Linux workloads on Power Systems, VMware’s ESXi hypervisor and the open source KVM hypervisor have not been able to meet some of the requirements of virtualized HANA setups that PowerVM was able to handle, such as guaranteeing performance when an LPAR “goes hot,” as he put it.
As is the case with database management systems in the enterprise in general, the same holds true with HANA, and according to SAP, somewhere on the order of 80 percent of the HANA systems running out there have less than 6 TB of main memory in which they are running HANA. “I think that sometimes IBM, Hewlett Packard Enterprise, and SGI, which is now part of HPE, focused too much on the big systems when talking about HANA,” says Moranta. “What we are seeing is that for many customers, the HANA database is 512 GB or some of them are 1 TB.” That said, it doesn’t hurt that the Power E950 has 16 TB and the Power E980 has 64 TB of memory capacity, with plenty of cores to chew on that data and lots of memory and I/O bandwidth.
This is why, in any given quarter, IBM has somewhere between 20 percent and 30 percent of SAP HANA system sales, and that is all Power Systems sales, and interestingly, somewhere between 20 percent and 30 percent of Power Systems revenues are driven by SAP HANA. These are rough estimates that various IBMers have said to us over the past several years, and Moranta says they are about right and he also agrees that the IBM slice of the HANA piece is growing. In fact, he says that there have been four consecutive years of growth of SAP HANA on Power Systems.
That was certainly part of IBM’s plan when it started selling SAP HANA on Power iron back in 2015, and as you will recall, we got the inside dope on IBM’s playbook for SAP HANA on Power back in May 2018 and IBM’s own positioning of Power Systems back in the fall of 2018. IBM always expected for HANA to drive a big portion of its Linux-on-Power Systems revenues, and it looks like that is happening.
Here is the thing. If SAP is putting Power E950s and Power E980s in its own cloud to support SAP HANA and also S/4HANA customers, that is a good indicator that there is customer demand for SAP HANA on Power. SAP is spending its own money. Similarly, we think that IBM will make SAP HANA available on the Power E880 and Power E980 servers that it has on the IBM Cloud public cloud, and it is not outside the realm of possibility that the other clouds – Google, Skytap, and Microsoft – also make HANA available on their cloudy Power iron, too.
We think it will take a long time for IBM i customers to make the move from Business Suite running atop the Db2 for i database to the S/4HANA suite running on HANA on Linux on Power. They may be the last holdouts in the SAP base, we think, perhaps using HANA as a sidecar query engine for Business Suite as HANA was originally conceived before SAP decided it wanted to have it all and march all Business Suite customers to HANA and also to encourage them to move to S/4HANA applications while they are at it.
We shall see how that transition all goes in the next decade. All we know is that the SAP software stack is helping both IBM i and overall Power Systems sales right now, and that is what counts.
RELATED STORIES
Where Does SAP On IBM i Go From Here?
SAP Sending Mixed Messages On ERP Platform Support
Inside IBM’s SAP HANA On Power Playbook
IBM i Shops Not In A Hurry For HANA
SAP HANA Now Available On IBM Power
HANA On Power Marches Toward GA
What HANA on Power Means to IBM i
SAP Hana For IBM Power Officially In The Works
IBM Power Systems Can Do Big Data Analytics, Too
IBM Schedules SAP On i Summit, SAP Looks To $30 Billion Goal
SAP HANA: Just a Sidecar to IBM i, For Now
HANA on Power? It May Be in the Cards
Cloudy And HANA In-Memory Apps Lift SAP In Q1
SAP On IBM i: The Best Alternative?