You are currently on IBM Systems Media’s archival website. Click here to view our new website.

Bookmark and Share

Recent Posts

DB2 for i and SAP Live Up to the Hype

August 21, 2013

By Ron Schmerbauch, the technical leader of the SAP on IBM i team

What do we know about new and shiny things? Whether it is the latest smartphone or the latest DB platform, we hear a lot of marketing hype, based on narrow specialized tests with grandiose promises. They are usually expensive.  Conversion may be required and it might not be painless. They still might be fragile and untested in the real world. In short, we may find out that the hype is not quite all it was made out to be and this is why enterprise IT architects pay special attention to integration and service levels, and veteran platforms continue to persist despite any niche newcomers.

Big Data and Analytics currently see a lot of this hype with in-memory databases being the latest rage, especially in the SAP market. Architects need to be mindful that although Big Data and Analytics are important, they are only one part of a complete SAP solution. It is crucial to choose a solid architecture as a foundation for an SAP solution across the board - something time proven with a track record of reliability in both OLTP and OLAP workloads. 

Perhaps this explains how SAP running on IBM i and DB2 for IBM i is now entering its 20th year of development. Yes, in case you didn’t know, IBM i runs SAP applications on Power Systems very well and has for quite a long time. Other databases have entered the SAP market since the 1990s, including a couple from SAP itself, yet DB2 on IBM i is still current with all of the latest SAP NetWeaver releases.  

Not only are SAP NetWeaver applications supported on IBM i, IBM i excels in both performance and TCO when it comes to SAP applications.  

For example, DB2 for IBM i is giving SAP’s shiny new HANA DB platform a run for its money in the latest SAP benchmark, BW-EML. Considering this workload was specifically architected in an attempt to showcase SAP HANA against traditional databases by putting a premium on complex ad-hoc queries with random selection criteria, IBM i shows extremely well at the official SAP scoreboard for the SAP BW-EML benchmark, setting the throughput record with the most Ad-Hoc Navigation Steps per hour.

These SAP benchmark results are spectacular when one considers that IBM i is running both DB2 for i and the SAP application server together within one box and within just one partition, saving on administration overhead – not to mention floor space, cooling and electrical power. IBM i was architected to support multiple business applications and multiple users at once. It’s quite typical for IBM i clients to leverage IBM i subsystems to run multiple SAP components in a single partition, and to take advantage of PowerVM to support multiple partitions on a single footprint. These architectural advantages of POWER and IBM i together provide an amazing TCO story for the SAP landscape as this study explains.

The native Single Level Storage (SLS) concept on IBM i is another helpful architectural feature when noting that SAP databases tend to grow in size at about 20-50% annually. Although SLS includes many other sophisticated capabilities, it delivers outstanding benefits for workloads of this type. IBM i will automatically benefit from larger memory capacity whenever physical memory is added to the system just as an in-memory DB platform would. But whenever necessary the IBM i SLS memory space expands seamlessly to disk ensuring that database size is not limited by physical memory.  With SSD and Flash Storage becoming more common, disk speeds get ever closer to memory speeds and the benefit of SLS is further amplified.

It is disappointing to see that a “modern” database like SAP HANA is perpetuating the x86 server sprawl problem, where “just buy another one” seems to be the answer to just about any question.  HANA must dedicate x86 resources just for its DB work, forcing the SAP application server to run elsewhere. It also must be sized to make sure the in-memory database never simply stops because the size of the database exceeds the capacity of the physical memory on the system. Is this really progress? Sprawl, oversizing? That’s just the first system; consider how much more is required for any sense of High Availability and imagine the impact to ROI.

In the benchmark configuration and the study above, we’ve shown that IBM i is able to manage and share resources in an architecturally elegant way to provide both great performance and TCO. These features enable clients running SAP on IBM i to spend time focusing on business challenges instead of worrying about when and where they need to plug in the next  x86 server to keep their infrastructure from toppling. I’d call that real business value, worthy of some hype.




Posted August 21, 2013| Permalink