IBM Systems Magazine, Mainframe - March/April 2009 - 46

Technical Corner If I were to offer any advice on future trends for the z/OS system programmer, it would be to embrace change and new technologies. consider gathering type 42 for Systems Managed Storage I/O. When identifying poor-performing SQL, run traces sparingly and only under a controlled interval of no longer than ﬁve minutes. Only use the following trace command to capture speciﬁc SQL: -START TRACE(MON) CLASS(1) IFCID(316,317,318) DEST(SMF) If I were to offer any advice on future trends for the z/OS system programmer, it would be to embrace change and new technologies. Combining the old with the new will be a good key to success as these job tasks encompass more than just another SMP/E installation. Jeff M. Sullivan is a software specialist SWAT team member on the data warehouse solutions on the System z team. He has more than 28 years working on IBM mainframes mostly as a z/OS and DB2 systems programmer/database administrator. Jeff can be reached at jeffsull@us.ibm.com. If you use a distributed BI application, keep a careful watch on the WLM service class associated with distributed data facility (DDF). Like many two- and three-tier architected applications, distributed DW applications will generate the SQL used and employ DB2 Connect* to communicate Table 2 Suggested Strategy with your DW tables and/or operational data store on DB2 for z/OS. Before changing anything BP0 – DB2 Catalog i n W L M , r e me mbe r t he law of u n i nte nde d BP1 – Small-sized Reference Tablespaces consequences, which states any purposeful action BP2 – Small-sized Reference Indexspaces can produce some unintended consequences. Pay attention to all service policy-performance BP3 to BP6, BP8 to BP9 – expansion/isolation for performance indexes while making only subtle changes to bottlenecks in BP1 and BP2 the policies. BP7 – Sort DSNDB07 Pick up a good Linux* reference. I say this BP10 to BP19 – Tablespace bufferpools not so muc h for data wa re hou si ng, but for f ut ure trends. Over t he years t here’ve been BP10 – Medium-sized Sequential Access subtle changes in the systems-programming BP11 – Medium-sized Random Access environment, with more emphasis on open-system BP12 – Large-sized Sequential Access applications. This is a suggestion to help you stay BP13 – Large-sized Random Access ahead of the trend. Consider making a cross BP14 to BP19 – expansion/isolation for performance bottlenecks reference of commands used wit h Linu x on BP20 to BP29 – Indexspace bufferpools System z or Interactive System Productivit y BP20 – Medium-sized Sequential Access Facility (ISPF) panels and commands. BP21 – Medium-sized Random Access BP22 – Large-sized Sequential Access A Shift in the System-Programmer Role BP23 – Large-sized Random Access on the Mainframe BP24 to BP29 – expansion/isolation for performance bottlenecks The days of the pure system programmer are sh if t i ng a nd for t he pa st sever a l yea r s t he systems-programmer function has been evolving into a more full-service role. We’re asked to not only install products, but also install on non-z/OS OSs. Additionally, we’re asked to not only provide initial conﬁguration settings, but also watch the overall performance for bottlenecks in memory, ce nt r a l e le c t ron ic comple x ( C EC) capac it y and I/O. 46 M A R C H/A PRIL 2009 References More information can be found in the IBM Redbooks publications on DB2 performance: “DB2 UDB for z/OS Version 8 Performance Topics” (SG246465, www.redbooks.ibm.com/abstracts/sg246465.html?Open) and “DB2 9 for z/OS Performance Topics” (SG247473, www.redbooks. ibm.com/abstracts/sg247473.html?Open). ibmsystemsmag.com/mainframe

http://www.redbooks.ibm.com/abstracts/sg246465.html?Open http://www.redbooks.ibm.com/abstracts/sg247473.html?Open http://www.redbooks.ibm.com/abstracts/sg247473.html?Open http://www.ibmsystemsmag.com/mainframe

IBM Systems Magazine, Mainframe - March/April 2009

Table of Contents for the Digital Edition of IBM Systems Magazine, Mainframe - March/April 2009