Cover image for Advanced POWER Virtualization on IBM eServer p5 Servers : Architecture and Performance Considerations.
Advanced POWER Virtualization on IBM eServer p5 Servers : Architecture and Performance Considerations.
Title:
Advanced POWER Virtualization on IBM eServer p5 Servers : Architecture and Performance Considerations.
Author:
Redbooks, IBM.
Personal Author:
Physical Description:
1 online resource (436 pages)
Contents:
Contents -- Notices -- Trademarks -- Preface -- The specialists who wrote this redbook -- Become a published author -- Comments welcome -- Part 1 Virtualization technology -- Chapter 1. Introduction -- 1.1 Performance tuning redefined -- 1.1.1 Understanding performance -- 1.1.2 Performance considerations -- Chapter 2. IBM POWER5 architecture -- 2.1 Introduction -- 2.2 Chip design -- 2.3 POWER5 enhancements -- 2.4 POWER5 instruction pipelines -- 2.4.1 Instruction fetching -- 2.4.2 Branch prediction -- 2.4.3 Instruction decoding and preprocessing -- 2.4.4 Group dispatch -- 2.4.5 Register renaming -- 2.4.6 Instruction execution -- 2.5 Caches -- 2.5.1 Level 2 (L2) cache -- 2.5.2 Level 3 (L3) cache -- 2.5.3 Summary of caches on POWER5 -- 2.5.4 Address translation resources -- 2.6 Timing facilities -- 2.7 Dynamic power management -- 2.8 Processor Utilization Resource Register (PURR) -- 2.9 Large POWER5 SMPs -- 2.10 Summary -- Chapter 3. Simultaneous multithreading -- 3.1 What is multithreading? -- 3.2 POWER5 simultaneous multithreading features -- 3.2.1 Dynamic switching of thread states -- 3.2.2 Snooze and snooze delay -- 3.3 Controlling priority of threads -- 3.3.1 Dynamic resource balancing (DRB) -- 3.3.2 Adjustable thread priorities -- 3.3.3 Thread priority implementation -- 3.4 Software considerations -- 3.4.1 Simultaneous multithreading aware scheduling -- 3.4.2 Thread priorities on AIX 5L V5.3 -- 3.4.3 Thread priorities on Linux -- 3.4.4 Cache effects -- 3.5 Simultaneous multithreading performance -- 3.5.1 Engineering and scientific applications -- 3.5.2 Simultaneous multithreading benchmarks -- 3.6 Summary -- Chapter 4. POWER Hypervisor -- 4.1 POWER Hypervisor implementation -- 4.1.1 POWER Hypervisor functions -- 4.1.2 Micro-Partitioning extensions -- 4.1.3 POWER Hypervisor design -- 4.2 Performance considerations.

Chapter 5. Micro-Partitioning -- 5.1 Partitioning on the IBM eServer p5 systems -- 5.2 Micro-Partitioning implementation -- 5.2.1 Virtual processor dispatching -- 5.2.2 Phantom interrupts -- 5.3 Performance considerations -- 5.3.1 Micro-Partitioning considerations -- 5.3.2 Locking considerations -- 5.3.3 Memory affinity considerations -- 5.3.4 Idle partition consideration -- 5.3.5 Application considerations in Micro-Partitioning -- 5.3.6 Micro-Partitioning planning guidelines -- 5.4 Summary -- Chapter 6. Virtual I/O -- 6.1 Introduction -- 6.2 POWER Hypervisor support for virtual I/O -- 6.2.1 Virtual I/O infrastructure -- 6.2.2 Types of connections -- 6.3 The IBM Virtual I/O Server -- 6.3.1 Providing high availability support -- 6.4 Virtual Serial Adapter (VSA) -- 6.5 Virtual Ethernet -- 6.5.1 Virtual LAN -- 6.5.2 Virtual Ethernet connections -- 6.5.3 Benefits of virtual Ethernet -- 6.5.4 Limitations and considerations -- 6.5.5 POWER Hypervisor switch implementation -- 6.5.6 Performance considerations -- 6.5.7 VLAN throughput at different processor entitlements -- 6.5.8 Comparing throughput of VLAN to physical Ethernet -- 6.5.9 Comparing CPU utilization -- 6.5.10 Comparing transaction rate and latency -- 6.5.11 VLAN performance -- 6.5.12 VLAN implementation guidelines -- 6.6 Shared Ethernet Adapter -- 6.6.1 Shared Ethernet Adapter performance -- 6.6.2 Request/response time and latency -- 6.7 Implementation guidelines -- 6.7.1 Guidelines for Shared Ethernet Adapter sizing -- 6.7.2 Guidelines for physical Ethernet sizing -- 6.7.3 Control of threading in the Shared Ethernet Adapter -- 6.8 Virtual SCSI -- 6.8.1 Client and server interaction -- 6.8.2 AIX 5L V5.3 device configuration for virtual SCSI -- 6.8.3 Interpartition communication -- 6.8.4 Disk considerations -- 6.8.5 Configuring for redundancy -- 6.8.6 Performance considerations.

6.8.7 Sizing a virtual SCSI server -- 6.9 Summary -- Part 2 Virtualization support and tuning -- Chapter 7. AIX 5L Version 5.3 operating system support -- 7.1 Introduction -- 7.1.1 Processors -- 7.1.2 Dynamic re-configuration -- 7.1.3 Existing performance commands enhancement -- 7.1.4 New performance commands -- 7.1.5 Paging space -- 7.1.6 Logical Volume Manager (LVM) -- 7.1.7 Virtual local area network (VLAN) -- 7.1.8 EtherChannel -- 7.1.9 Partition Load Manager -- Chapter 8. POWER5 system performance -- 8.1 Performance commands -- 8.1.1 lparstat command -- 8.1.2 mpstat command -- 8.1.3 vmstat command -- 8.1.4 iostat command -- 8.1.5 sar command -- 8.1.6 topas command -- 8.1.7 xmperf command -- 8.2 Performance tuning approach -- 8.2.1 Global performance analysis -- 8.2.2 CPU analysis -- 8.2.3 Memory analysis -- 8.2.4 Disk I/O analysis -- 8.2.5 Network I/O analysis -- Chapter 9. Application tuning -- 9.1 Performance bottlenecks identification -- 9.1.1 Time commands, time utilities, and time routines -- 9.2 Tuning applications using only the compiler -- 9.2.1 Compiler brief overview -- 9.2.2 Most commonly used flags -- 9.2.3 Compiler directives for performance -- 9.2.4 POWER5 compiler features -- 9.3 Profiling applications -- 9.3.1 Hardware performance monitor -- 9.3.2 Profiling utilities -- 9.4 Memory management -- 9.5 Optimization of critical sections in the code -- 9.5.1 General rules for optimization strategies -- 9.5.2 Array optimization -- 9.5.3 Loop optimization -- 9.6 Optimized libraries -- 9.6.1 MASS Library -- 9.6.2 ESSL library -- 9.7 Parallel programming general concepts -- Chapter 10. Partition Load Manager -- 10.1 When and how should I use Partition Load Manager? -- 10.1.1 Partition Load Manager and other load-balancing tools -- 10.1.2 When to use Partition Load Manager -- 10.1.3 How to deploy Partition Load Manager.

10.2 More about Partition Load Manager installation and setup -- 10.2.1 Overview of Partition Load Manager behavior -- 10.2.2 Management versus monitoring modes -- 10.2.3 Configuration file and tunables -- 10.3 Managing and monitoring with Partition Load Manager -- 10.3.1 Managing multiple partitions -- 10.3.2 Extra tips about the xlplm command -- 10.3.3 Examples of Partition Load Manager commands output -- 10.4 Partition Load Manager performance impact -- 10.4.1 Partition Load Manager resource requirements -- 10.4.2 Partition Load Manager impact on managed partitions -- Related publications -- IBM Redbooks -- Other publications -- Online resources -- How to get IBM Redbooks -- Help from IBM -- Index -- Back cover.
Local Note:
Electronic reproduction. Ann Arbor, Michigan : ProQuest Ebook Central, 2017. Available via World Wide Web. Access may be limited to ProQuest Ebook Central affiliated libraries.
Electronic Access:
Click to View
Holds: Copies: