SAP on DB2 pureScale
In case you are new to DB2 pureScale, let me start with some recommended reading here:
- For DB2 pureScale in general, I think DB2 pureScale Overview from Matt Huras and DB2 pureScale Overview and Deep Dive from Brian Olynyk are very good overview presentations, especially if you are looking for the technical facts.
- The DB2 pureScale documentation is available in the IBM DB2 pureScale Feature Information Center.
- Regarding SAP on DB2 pureScale you will certainly find the presentation DB2 pureScale for SAP applications very helpful.
- And finally you can find the details how to install a SAP system on DB2 pureScale in our new extensive guide Running an SAP System on DB2 pureScale.
As usual SAP and IBM worked closely together to bring this new technology to its customers as soon as possible. DB2 pureScale is modeled after the Parallel Sysplex cluster on z/OS, so we tried to benefit from the experiences made by running SAP on the Parallel Sysplex as much as possible. Currently, there are three areas with SAP specific enhancements with respect to DB2 pureScale:
- Probably the most important point is the introduction of the client affinity concept for SAP systems running on DB2 pureScale. Basically, it says that all workprocesses of every SAP application server connect against one DB2 pureScale member. The other way around, a pureScale member can serve of course many SAP applications servers. The concept further defines which DB2 member will be contacted first and what happens in case of a failover. This concept is different from the standard DB2 pureScale workload based connection mechanism. It is tailored to the needs of SAP application servers and ensures maximal scalability in a SAP environment.
- Monitoring extension for the SAP DBACOCKPIT deal with specifics of DB2 pureScale, e.g. with the new Cluster Caching Facility and the new locking concepts. See SAP note 1409540 for how to enhance you existing DBACOCKPIT with the new monitoring functionality for DB2 pureScale.
- A special setup on the so called update tables (VBHDR, VBMOD, VBDATA) avoids locking situations. Basically, every SAP applications server gets its own set of these update tables. This concepts is known as update table multiplexing.
All these concepts are described in more detail in the guide Running an SAP System on DB2 pureScale mentioned above.
Registration for the DB2 pureScale pilot program is open, the details are in SAP note 1419433. Registered customers will have full support from IBM and SAP for their SAP systems running on DB2 pureScale. Initially we use a special build of Fix Pack 2 – available for AIX and Linux – for the pilot program.