Front cover -- Contents -- Notices -- Trademarks -- Preface -- The team that wrote this redbook -- Become a published author -- Comments welcome -- Part 1 Introduction -- Chapter 1. Performance overview -- 1.1 Performance expectations -- 1.1.1 System workload -- 1.1.2 Performance objectives -- 1.1.3 Program execution model -- 1.1.4 System tuning -- 1.2 Introduction to the performance tuning process -- 1.2.1 Performance management phases -- Chapter 2. Performance analysis and tuning -- 2.1 CPU performance -- 2.1.1 Processes and threads -- 2.1.2 SMP performance -- 2.1.3 Initial advice for monitoring CPU -- 2.2 Memory overview -- 2.2.1 Virtual memory manager (VMM) overview -- 2.2.2 Paging space overview -- 2.3 Disk I/O performance -- 2.3.1 Initial advice -- 2.3.2 Disk subsystem design approach -- 2.3.3 Bandwidth-related performance considerations -- 2.3.4 Disk design -- 2.3.5 Logical Volume Manager concepts -- 2.4 Network performance -- 2.4.1 Initial advice -- 2.4.2 TCP/IP protocol -- 2.4.3 Network tunables -- Part 2 Performance tools -- Chapter 3. General performance monitoring tools -- 3.1 The topas command -- 3.1.1 Topas syntax -- 3.1.2 Basic topas output -- 3.1.3 Partition statistics -- 3.2 The jtopas utility -- 3.2.1 The jtopas configuration file -- 3.2.2 The info section for the jtopas tool -- 3.2.3 The jtopas consoles -- 3.2.4 The jtopas playback tool -- 3.3 The perfpmr utility -- 3.3.1 Information about measurement and sampling -- 3.3.2 Building and submitting a test case -- 3.3.3 Examples for perfpmr -- 3.4 Performance Diagnostic Tool (PDT) -- 3.4.1 Examples for PDT -- 3.4.2 Using reports generated by PDT -- 3.4.3 Running PDT collection manually -- 3.5 The curt command -- 3.5.1 Information about measurement and sampling -- 3.5.2 Examples for curt -- 3.5.3 Overview of the reports generated by curt -- 3.5.4 The default report.

3.6 The splat command -- 3.6.1 splat syntax -- 3.6.2 Information about measurement and sampling -- 3.6.3 The execution, trace, and analysis intervals -- 3.6.4 Trace discontinuities -- 3.6.5 Address-to-name resolution in splat -- 3.6.6 splat examples -- 3.7 The trace, trcnm, and trcrpt commands -- 3.7.1 The trace command -- 3.7.2 Information about measurement and sampling -- 3.7.3 How to start and stop trace -- 3.7.4 Running trace interactively -- 3.7.5 Running trace asynchronously -- 3.7.6 Running trace on an entire system for 10 seconds -- 3.7.7 Tracing a command -- 3.7.8 Tracing using one set of buffers per CPU -- 3.7.9 Examples for trace -- 3.7.10 The trcnm command -- 3.7.11 Examples for trcnm -- 3.7.12 The trcrpt command -- 3.7.13 Examples for trcrpt -- Chapter 4. CPU analysis and tuning -- 4.1 CPU overview -- 4.1.1 Performance considerations with POWER4-based systems -- 4.1.2 Performance considerations with POWER5-based systems -- 4.2 CPU monitoring -- 4.2.1 The lparstat command -- 4.2.2 The mpstat command -- 4.2.3 The procmon tool -- 4.2.4 The topas command -- 4.2.5 The sar command -- 4.2.6 The iostat command -- 4.2.7 The vmstat command -- 4.2.8 The ps command -- 4.2.9 The trace tool -- 4.2.10 The curt command -- 4.2.11 The splat command -- 4.2.12 The truss command -- 4.2.13 The gprof command -- 4.2.14 The pprof command -- 4.2.15 The prof command -- 4.2.16 The tprof command -- 4.2.17 The time command -- 4.2.18 The timex command -- 4.3 CPU related tuning tools and techniques -- 4.3.1 The smtctl command -- 4.3.2 The bindintcpu command -- 4.3.3 The bindprocessor command -- 4.3.4 The schedo command -- 4.3.5 The nice command -- 4.3.6 The renice command -- 4.4 CPU summary -- 4.4.1 Other useful commands for CPU monitoring -- Chapter 5. Memory analysis and tuning -- 5.1 Memory monitoring -- 5.1.1 The ps command -- 5.1.2 The sar command.

5.1.3 The svmon command -- 5.1.4 The topas monitoring tool -- 5.1.5 The vmstat command -- 5.2 Memory tuning -- 5.2.1 The vmo command -- 5.2.2 Paging space thresholds tuning -- 5.3 Memory summary -- 5.3.1 Other useful commands for memory performance -- 5.3.2 Paging space commands -- Chapter 6. Network performance -- 6.1 Network overview -- 6.1.1 The maxmbuf tunable -- 6.2 Hardware considerations -- 6.2.1 Firmware levels -- 6.2.2 Media speed considerations -- 6.2.3 MTU size -- 6.3 Network monitoring -- 6.3.1 Creating network load -- 6.4 Network monitoring commands -- 6.4.1 The entstat command -- 6.4.2 The netstat command -- 6.4.3 The pmtu command -- 6.5 Network packet tracing tools -- 6.5.1 The iptrace command -- 6.5.2 The ipreport command -- 6.5.3 The ipfilter command -- 6.5.4 The netpmon command -- 6.5.5 The trpt command -- 6.6 NFS related performance commands -- 6.6.1 The nfsstat command -- 6.7 Network tuning commands -- 6.7.1 The no command -- 6.7.2 The Interface Specific Network Options (ISNO) -- 6.7.3 The nfso command -- Chapter 7. Storage analysis and tuning -- 7.1 Data placement and design -- 7.1.1 AIX I/O stack -- 7.1.2 Physical disk and disk subsystem -- 7.1.3 Device drivers and adapters -- 7.1.4 Volume groups and logical volumes -- 7.1.5 VMM and direct I/O -- 7.1.6 JFS/JFS2 file systems -- 7.2 Monitoring -- 7.2.1 The iostat command -- 7.2.2 The filemon command -- 7.2.3 The fileplace command -- 7.2.4 The lslv, lspv, and lsvg commands -- 7.2.5 The lvmstat command -- 7.2.6 The sar -d command -- 7.3 Tuning -- 7.3.1 The lsdev, rmdev and mkdev commands -- 7.3.2 The lscfg, lsattr, and chdev commands -- 7.3.3 The ioo command -- 7.3.4 The lvmo command -- 7.3.5 The vmo command -- Part 3 Case studies and miscellaneous tools -- Chapter 8. Case studies -- 8.1 Case study: NIM server -- 8.1.1 Setting up the environment.

8.1.2 Monitoring NIM master using topas -- 8.1.3 Upgrading NIM environment to Gbit Ethernet -- 8.1.4 Upgrading the disk storage -- 8.1.5 Real workload with spread file system -- 8.1.6 Summary -- 8.2 POWER5 case study -- 8.2.1 POWER5 introduction -- 8.2.2 High CPU -- 8.2.3 Evaluation -- Chapter 9. Miscellaneous tools -- 9.1 Workload manager monitoring (WLM) -- 9.1.1 Overview -- 9.1.2 WLM concepts -- 9.1.3 Administering WLM -- 9.1.4 WLM performance tools -- 9.2 Partition load manager (PLM) -- 9.2.1 PLM introduction -- 9.2.2 Memory management -- 9.2.3 Processor management -- 9.3 A comparison of WLM and PLM -- 9.4 Resource monitoring and control (RMC) -- 9.4.1 RMC commands -- 9.4.2 Information about measurement and sampling -- 9.4.3 Verifying RMC facilities -- 9.4.4 Examples using RMC -- Chapter 10. Performance monitoring APIs -- 10.1 The performance status (Perfstat) API -- 10.1.1 Compiling and linking -- 10.1.2 Changing history of perfstat API -- 10.1.3 Subroutines -- 10.2 System Performance Measurement Interface -- 10.2.1 Compiling and linking -- 10.2.2 Terms and concepts for SPMI -- 10.2.3 Subroutines -- 10.2.4 Basic layout of SPMI program -- 10.2.5 SPMI examples -- 10.3 Performance Monitor API -- 10.3.1 Performance Monitor data access -- 10.3.2 Compiling and linking -- 10.3.3 Subroutines -- 10.3.4 PM API examples -- 10.3.5 PMAPI M:N pthreads support -- 10.4 Miscellaneous performance monitoring subroutines -- 10.4.1 Compiling and linking -- 10.4.2 Subroutines -- 10.4.3 Combined example -- Appendix A. Source code -- perfstat_dump_all.c -- perfstat_dude.c -- spmi_dude.c -- spmi_data.c -- spmi_file.c -- Spmi_traverse.c -- dudestat.c -- Appendix B. Trace hooks -- AIX 5L trace hooks -- Abbreviations and acronyms -- Related publications -- IBM Redbooks -- Other publications -- Online resources -- How to get IBM Redbooks -- Help from IBM -- Index.

Back cover.