HP-UX 11i Tuning and Performance,9780131433496

HP-UX 11i Tuning and Performance

by ; ;
Edition: 2nd
ISBN13: 9780131433496
ISBN10: 0131433490
Format: Paperback
Pub. Date: 2004-03-24
Publisher(s): Prentice Hall
Availability: This title is currently not available.
List Price: $59.99

Rent Book

Select for Price
Add to Cart
There was a problem. Please try again later.

New Book

We're Sorry
Sold Out

Used Book

We're Sorry
Sold Out

eBook

We're Sorry
Not Available

Summary

Maximize the performance and value of any HP-UX 11i system.

Author Biography

Robert F. Sauers is HP's Chief Architect for Mission Critical and Disaster Tolerant Solutions Chris P. Ruemmler is an HP Software Engineer responsible for HP-UX performance Peter S. Weygant is Principal Consultant with PSW Consulting, Inc.

Table of Contents

About the Authors xxiii
Acknowledgments xxv
Chapter 1 Introduction to Performance Management 1(12)
1.1 Application Developer's Perspective
1(2)
1.1.1 Defining the Application
2(1)
1.1.2 Determining Application Specifications
2(1)
1.1.3 Designing Application Components
2(1)
1.1.4 Developing the Application Codes
2(1)
1.1.5 Testing, Tuning, and Debugging
2(1)
1.1.6 Deploying the System and/or Application
3(1)
1.1.7 System and Application Maintenance
3(1)
1.2 System Administrator's Perspective
3(1)
1.2.1 Making the System Available to the Users
3(1)
1.2.2 Monitoring the Usage of the System
3(1)
1.2.3 Maintaining a Certain Level of System Performance
3(1)
1.2.4 Planning for Future Processing Needs
4(1)
1.3 Total System Resource Perspective
4(1)
1.4 Rules of Performance Tuning
5(8)
Chapter 2 Performance Management Tasks 13(6)
2.1 Workload Estimation
14(1)
2.2 Benchmarking
15(1)
2.3 Performance Characterization
15(1)
2.4 Performance Forecasting or Prediction
15(1)
2.5 Application Optimization
16(1)
2.6 Capacity Planning
16(1)
2.7 Performance Problem Resolution
17(2)
Chapter 3 A Performance Management Methodology 19(20)
3.1 Assessment
19(7)
3.1.1 System Configuration
20(1)
3.1.2 Application Design
21(1)
3.1.3 Performance Expectations
21(4)
3.1.4 Known Peak Periods
25(1)
3.1.5 Sudden Changes
25(1)
3.1.6 Duration of the Problem
25(1)
3.1.7 Understanding the Options
26(1)
3.1.8 Understanding the Politics
26(1)
3.2 Measurement
26(5)
3.2.1 Tool Availability
27(1)
3.2.2 Purpose of the Measurement: Baseline versus Crisis
27(1)
3.2.3 Duration and Interval of the Measurement
28(1)
3.2.4 Particular Metric Needs
29(1)
3.2.5 Metric Documentation
29(1)
3.2.6 Metric Accuracy
29(1)
3.2.7 Saturation of Certain System Resources
30(1)
3.2.8 Relationship Between the Metric and the Application
30(1)
3.2.9 Qualitative versus Quantitative Measurements
30(1)
3.2.10 Summary
30(1)
3.3 Interpretation and Analysis
31(2)
3.3.1 Multi-User versus Workstation
31(1)
3.3.2 Interactive versus Batch, Compute-Intensive versus I/O-Intensive
32(1)
3.3.3 Application Architecture
32(1)
3.3.4 Performance of the CPU
32(1)
3.3.5 Type of Disks
32(1)
3.3.6 Type of Networks
33(1)
3.4 Identifying Bottlenecks
33(1)
3.4.1 Resource Saturation
33(1)
3.4.2 Growing Resource Queue
33(1)
3.4.3 Resource Starvation
34(1)
3.4.4 Unsatisfactory Response Time
34(1)
3.4.5 Bottleneck Summary
34(1)
3.5 Tuning or Upgrading
34(5)
3.5.1 Determine Whether There Is a Particular Bottleneck
35(1)
3.5.2 Devise a Simple Test for Measuring Results
35(1)
3.5.3 Do Not Tune Randomly
35(1)
3.5.4 Use Heuristics and Logic
35(1)
3.5.5 Look for Simple Causes
35(1)
3.5.6 Develop an Action Plan
36(1)
3.5.7 Change Only One Thing at a Time!
36(1)
3.5.8 Prioritize Goals
36(1)
3.5.9 Understand the Limits of the Application's Architecture
36(1)
3.5.10 Know When to Stop
36(3)
Chapter 4 Kernel Instrumentation and Performance Metrics 39(12)
4.1 Approaches to Measuring
39(7)
4.1.1 External Hardware Monitors
40(1)
4.1.2 Internal Hardware Monitors
40(1)
4.1.3 Internal Software Monitors
41(5)
4.2 Kernel Instrumentation and Measurement Interface
46(2)
4.2.1 Acceptable Overhead
47(1)
4.3 Performance Metrics Categories
48(1)
4.3.1 Categories of Metrics by Scope
48(1)
4.3.2 Categories of Metrics by Type of Data
49(1)
4.4 Summary
49(2)
Chapter 5 Survey of Unix Performance Tools 51(56)
5.1 Multipurpose Diagnostic Tools
54(23)
5.1.1 GlancePlus/UX
54(5)
5.1.2 GlancePlus Motif
59(7)
5.1.3 GlancePlus Motif Adviser
66(1)
5.1.4 Advantages and Disadvantages of GlancePlus Motif
67(1)
5.1.5 sar (System Activity Reporter)
68(6)
5.1.6 time
74(1)
5.1.7 timex
75(1)
5.1.8 uptime
76(1)
5.2 CPU Diagnostic Tools
77(4)
5.2.1 ps
77(1)
5.2.2 top
78(1)
5.2.3 xload
79(1)
5.2.4 Machinfo
79(2)
5.3 Disk Diagnostic Tools
81(4)
5.3.1 bdf and df
81(2)
5.3.2 ioscan
83(1)
5.3.3 iostat
84(1)
5.4 Memory Diagnostic Tools
85(6)
5.4.1 size
85(1)
5.4.2 swapinfo
86(1)
5.4.3 vmstat
87(2)
5.4.4 ipcs
89(1)
5.4.5 chatr
90(1)
5.5 Performance Characterization and Prediction Tools
91(6)
5.5.1 MeasureWare
91(3)
5.5.2 PerfView
94(2)
5.5.3 PerfView/Analyzer
96(1)
5.5.4 PerfView/Monitor
97(1)
5.5.5 PerfView/Planner
97(1)
5.5.6 sar and sadc
97(1)
5.6 Process Accounting Tools
97(1)
5.6.1 acct
98(1)
5.6.2 diskusg
98(1)
5.7 Application Optimization Tools
98(1)
5.7.1 Caliper
98(1)
5.7.2 CXperf
98(1)
5.7.3 dpat (Distributed Performance Analysis Tool)
98(1)
5.7.4 hpc (Histogram Program Counter)
99(1)
5.7.5 gprof/gprof++
99(1)
5.7.6 prof
99(1)
5.7.7 Puma
99(1)
5.7.8 TTV (Thread Trace Visualizer)
99(1)
5.8 Network Diagnostic Tools
99(2)
5.8.1 netstat
99(2)
5.8.2 nettune (HP-UX 10.x) and ndd (HP-UX 11.0 and Later)
101(1)
5.8.3 nfsstat
101(1)
5.9 Resource Management Tools
101(6)
5.9.1 Process Resource Manager (PRM)
103(1)
5.9.2 Workload Manager (WLM)
104(2)
5.9.3 Capacity Planning Tools
106(1)
Chapter 6 Hardware Performance Issues 107(62)
6.1 Terminology
107(9)
6.1.1 Central Processing Unit
108(1)
6.1.2 Main Memory
108(1)
6.1.3 Virtual Memory
108(1)
6.1.4 Cache Memory
108(1)
6.1.5 Translation Lookaside Buffers
109(1)
6.1.6 Branch Prediction
110(5)
6.1.7 Input/Output Devices (I/O Devices)
115(1)
6.1.8 Buses
115(1)
6.1.9 Graphics Devices
115(1)
6.2 Processor Characteristics
116(18)
6.2.1 Instruction Types
116(1)
6.2.2 Out-of-Order versus In-Order Execution
117(1)
6.2.3 Scalar and Superscalar Processors
118(2)
6.2.4 CPU Pipelines
120(2)
6.2.5 Problems with Pipelined Architectures
122(2)
6.2.6 EPIC Features
124(4)
6.2.7 Vectorization
128(1)
6.2.8 Summary of processors
128(6)
6.3 Multi-Processing
134(2)
6.3.1 Master/Slave Implementation
135(1)
6.3.2 Asymmetric Implementation
135(1)
6.3.3 Symmetric Multi-Processing Implementation
135(1)
6.3.4 Massively Parallel Processing Implementation
135(1)
6.3.5 Parallelization
136(1)
6.4 Cache Memory Performance Issues
136(6)
6.4.1 Cache Designs
136(2)
6.4.2 Visible Cache Problems
138(3)
6.4.3 Cache Coherence
141(1)
6.5 Main Memory Performance Issues
142(8)
6.5.1 Physical Memory
142(1)
6.5.2 Virtual Memory
143(4)
6.5.3 TLB Designs
147(1)
6.5.4 Visible TLB Performance Problems
148(2)
6.6 I/O Performance Issues
150(2)
6.6.1 I/O Backplane Types
150(1)
6.6.2 I/O card types
151(1)
6.6.3 I/O Problem Solving
152(1)
6.7 PA-RISC Systems
152(12)
6.7.1 Early Models
152(1)
6.7.2 700 Series
152(1)
6.7.3 890, T500, T520
152(1)
6.7.4 T600
153(1)
6.7.5 E-Series
153(1)
6.7.6 K-Series
153(3)
6.7.7 V-Series
156(1)
6.7.8 D-Series and R-Series
156(1)
6.7.9 N-Series (rp7400)
157(3)
6.7.10 L-series (rp54xx)
160(1)
6.7.11 Superdome
160(2)
6.7.12 The rp8400 Server
162(1)
6.7.13 The rp7405 and rp7410 Servers
163(1)
6.8 Itanium® Systems
164(5)
6.8.1 rx4650 server
165(1)
6.8.2 rx5670 and rx2600
165(1)
6.8.3 Integrity Superdome
166(1)
6.8.4 rx8620
167(1)
6.8.5 rx7620
167(2)
Chapter 7 CPU Bottlenecks 169(34)
7.1 Processes and Threads
170(5)
7.1.1 1x1 Threads
173(1)
7.1.2 MxN Threads
173(2)
7.1.3 Conclusions and recommendations
175(1)
7.2 Scheduling
175(11)
7.2.1 Time-share Scheduling
175(3)
7.2.2 Real-Time Scheduling
178(1)
7.2.3 Process Resource Manager (PRM) and CPU Resource Allocation
179(3)
7.2.4 Workload Manager (WLM)
182(1)
7.2.5 Context Switches
182(1)
7.2.6 SMP Scheduling Issues
183(3)
7.3 System Partitioning
186(7)
7.3.1 Hardware Partitioning
186(1)
7.3.2 Virtual Partitions (vPars)
187(1)
7.3.3 Processor Sets
188(1)
7.3.4 Processor Affinity
189(4)
7.4 Traps and Protection Violations
193(1)
7.5 Global CPU Metrics
194(1)
7.5.1 Global CPU Saturation Metrics
194(1)
7.5.2 Global CPU Queue Metrics: Run Queue vs. Load Average
195(1)
7.5.3 Other Global CPU Metrics
195(1)
7.5.4 Per-Process CPU Metrics
195(1)
7.6 Typical Metric Values
195(2)
7.7 Symptoms of a CPU Bottleneck
197(2)
7.7.1 Saturation
197(1)
7.7.2 Expensive CPU Utilization
197(1)
7.7.3 Expensive System Calls
198(1)
7.8 CPU Use and Performance Tools
199(1)
7.9 Tuning CPU Bottlenecks
200(3)
7.9.1 Hardware Solutions
200(1)
7.9.2 Software Solutions
201(1)
7.9.3 Application Optimization
201(1)
7.9.4 Tuning CPU-Related Parameters
202(1)
Chapter 8 Memory Bottlenecks 203(50)
8.1 Virtual Address Space
204(14)
8.1.1 PA-RISC Virtual Address Space Layout
204(4)
8.1.2 IPF Virtual Address Space Layout
208(3)
8.1.3 Modifying the Default Virtual Address Space Layout
211(3)
8.1.4 Adaptive Address Space
214(2)
8.1.5 Shared Memory Windows
216(1)
8.1.6 Page Protection
217(1)
8.2 Memory Organization
218(9)
8.2.1 Variable Page Sizes
218(4)
8.2.2 Cell Local Memory
222(5)
8.3 Paging Control
227(5)
8.3.1 Paging
227(1)
8.3.2 Operation of vhand
228(1)
8.3.3 Swapping
228(1)
8.3.4 Deactivation
229(1)
8.3.5 Memory Allocation Management with PRM
229(2)
8.3.6 Locking Memory
231(1)
8.3.7 Sticky Bit
231(1)
8.3.8 Serialization
231(1)
8.4 File Access
232(3)
8.4.1 Dynamic Buffer Cache
232(1)
8.4.2 Memory-Mapped Files
233(1)
8.4.3 Memory-Mapped Semaphores
234(1)
8.5 Process and Thread Execution
235(1)
8.5.1 fork and vfork
236(1)
8.5.2 pthread_create
236(1)
8.6 Malloc'ed Memory
236(2)
8.6.1 Mallopt
237(1)
8.6.2 Environment Options
238(1)
8.6.3 General Malloc Issues
238(1)
8.7 System V Shared Memory
238(2)
8.7.1 Allocating Shared Memory
238(1)
8.7.2 Attaching to a Shared Memory Segment
239(1)
8.7.3 Modifying Shared Memory Attributes
239(1)
8.7.4 Kernel Tuneable Parameters for Shared Memory
240(1)
8.8 Shared Libraries
240(1)
8.9 Memory Management Policies
240(9)
8.9.1 Regions and Pregions
241(1)
8.9.2 Thresholds and Policies
242(1)
8.9.3 Values for Memory Management Parameters
242(1)
8.10 Sizing Memory and the Swap Area
242(2)
8.10.1 Sizing the Swap Area
243(1)
8.10.2 Sizing Memory
243(1)
8.11 Memory Metrics
244(2)
8.11.1 Global Memory Saturation Metrics
244(1)
8.11.2 Global Memory Queue Metrics
244(1)
8.11.3 Other Global Memory Metrics
244(1)
8.11.4 Per-Process Memory Metrics
245(1)
8.11.5 Typical Metric Values
245(1)
8.12 Types of Memory Management Bottlenecks
246(1)
8.13 Expensive System Calls
246(1)
8.14 Tuning Memory Bottlenecks
247(2)
8.14.1 Hardware Solutions
247(1)
8.14.2 Software Solutions
247(1)
8.14.3 Application Optimization
248(1)
8.15 Memory-Related Tuneable Parameters
249(4)
8.15.1 Logical View of Physical Memory Utilization
251(2)
Chapter 9 Disk Bottlenecks 253(64)
9.1 Disk Hardware Descriptions
254(11)
9.1.1 Interface Cards (Host Bus Adapters or HBAs)
254(2)
9.1.2 Disk Mechanisms
256(2)
9.1.3 Disk Arrays
258(6)
9.1.4 HP Storage Works Disk Array XP-series
264(1)
9.1.5 Tape Drives
264(1)
9.2 Review of Disk I/O Concepts
265(7)
9.2.1 Disk Access Methods
265(1)
9.2.2 Unbuffered (Raw) UO
266(4)
9.2.3 Buffered UO
270(1)
9.2.4 Direct Files and Memory-Mapped Files
270(2)
9.2.5 Memory-Mapped I/O
272(1)
9.3 Logical Volume Manager Concepts
272(7)
9.3.1 LVM Striping
273(1)
9.3.2 LVM Mirroring
274(2)
9.3.3 Bad Block Relocation
276(1)
9.3.4 Read-Only Volume Groups
277(1)
9.3.5 Note on Mixing RAID Parity Hardware Striping and LVM Striping
277(1)
9.3.6 PV Links
278(1)
9.4 VxVM Volume Manager
279(1)
9.4.1 Performance Features
279(1)
9.4.2 Performance Problems
279(1)
9.4.3 Performance Measurement
280(1)
9.5 Shared-Bus Access to Disks
280(3)
9.5.1 Single-Initiator Configurations
280(1)
9.5.2 Multi-Initiator Configurations
280(1)
9.5.3 Performance Load on Multi-Initiator Buses
280(2)
9.5.4 Disk Drive Write Caches
282(1)
9.6 File Systems and the Kernel
283(4)
9.6.1 I/O-Related Kernel Tables
283(1)
9.6.2 File System Limits
284(1)
9.6.3 Buffer Cache
285(1)
9.6.4 Fixed Buffer Cache
286(1)
9.6.5 Dynamic Buffer Cache
286(1)
9.6.6 Maxusers, Maxfiles and Other Kernel Parameters
287(1)
9.7 File System Types
287(15)
9.7.1 HFS
287(5)
9.7.2 JFS
292(7)
9.7.3 OnlineJFS
299(1)
9.7.4 Quick I/O
300(1)
9.7.5 QuickLog
301(1)
9.7.6 Comparison of HFS and JFS
302(1)
9.8 Disk Metrics
302(4)
9.8.1 Global Disk Saturation Metrics
302(1)
9.8.2 Global Disk Queue Metrics
303(1)
9.8.3 Other Global Disk Metrics
303(2)
9.8.4 Per-Process Disk Metrics
305(1)
9.8.5 Typical Metric Values
305(1)
9.9 Types of Disk Bottlenecks
306(1)
9.10 Expensive System Calls
306(1)
9.11 Tuning Disk Bottlenecks
306(2)
9.11.1 Hardware Solutions
306(1)
9.11.2 Configuration Solutions
306(1)
9.11.3 File System Solutions
307(1)
9.11.4 Application Solutions
307(1)
9.12 Database Issues
308(2)
9.12.1 Use Raw Volumes
308(1)
9.12.2 Data Placement Recommendations
309(1)
9.12.3 Review and Tune Database Parameters
309(1)
9.13 Disk-Related Tuneable Parameters
310(7)
9.13.1 bufpages and nbuf
310(1)
9.13.2 dbc_max_pct and dbc_min_pct
311(1)
9.13.3 create_fastlinks
311(1)
9.13.4 nfile
311(1)
9.13.5 ninode
311(1)
9.13.6 max_async_ports
311(1)
9.13.7 maxvgs
312(1)
9.13.8 maxfiles
312(1)
9.13.9 maxfiles_lim, and no_lvm_disks
312(1)
9.13.10 scsi_max_qdepth
312(1)
9.13.11 HFS-Specific Tuneable Parameters
313(1)
9.13.12 JFS-Specific Tuneable Parameters
314(3)
Chapter 10 Network Bottlenecks 317(40)
10.1 Networking Hardware Descriptions
317(5)
10.1.1 Networking Infrastructure Devices
318(1)
10.1.2 Ethernet
319(1)
10.1.3 FDDI
319(1)
10.1.4 ATM
320(1)
10.1.5 Hyperfabric
321(1)
10.2 Review of Networking Concepts
322(7)
10.2.1 Interface Driver and DLPI
322(1)
10.2.2 IP
323(5)
10.2.3 TCP
328(1)
10.2.4 UDP
328(1)
10.2.5 Sockets
329(1)
10.2.6 Auto Port Aggregation
329(1)
10.2.7 Streams
329(1)
10.2.8 Checksum Offload
329(1)
10.3 Networking Access
329(4)
10.3.1 Ports
329(1)
10.3.2 FTP
330(1)
10.3.3 Telnet
330(1)
10.3.4 Rlogin
330(1)
10.3.5 Name Resolution
330(2)
10.3.6 World Wide Web
332(1)
10.4 Networked File System
333(6)
10.4.1 General NFS issues
333(1)
10.4.2 Client-Side Issues
334(1)
10.4.3 NFS Server-Side Issues
334(1)
10.4.4 UDP Versus TCP Issues for NFS
335(1)
10.4.5 NFS Options
336(1)
10.4.6 Measuring NFS Performance
337(1)
10.4.7 Local File System Usage With NFS
337(1)
10.4.8 Other NFS-Related Processes
337(2)
10.5 Clusters
339(2)
10.5.1 Heartbeat Networks in a Cluster
339(1)
10.5.2 Client Networks in a Cluster
340(1)
10.5.3 SGeRAC Lock Networks
340(1)
10.6 Network Metrics
341(6)
10.6.1 Global Network Saturation Metrics
341(1)
10.6.2 Other Global Network Metrics
341(1)
10.6.3 Network Adapter Metrics
342(1)
10.6.4 Per-Process Network Metrics
342(3092)
10.6.5 netstat Command
3434
10.7 Types of Network Bottlenecks
347(1)
10.8 Expensive System Calls
347(1)
10.9 Tuning Network Bottlenecks
348(9)
10.9.1 Hardware Solutions
348(1)
10.9.2 Configuration Solutions
348(1)
10.9.3 NFS File System Solutions
348(1)
10.9.4 Application Solutions
348(1)
10.10 Network-Related Tuneable Parameters
349(1)
10.10.1 Kernel Tuneable Parameters for Networking
349(1)
10.10.2 ndd
350(1)
10.10.3 lanadmin
351(1)
10.11 Web Server Tuning Issues
352(1)
10.11.1 Hardware Configuration
352(1)
10.11.2 ndd tuneables
352(1)
10.11.3 HP-UX Tuneables
353(1)
10.11.4 File System Tuning
353(1)
10.11.5 Networking Card Tuning
354(1)
10.11.6 Zeus Web Server Parameters
354(1)
10.12 Database Server Tuning Issues
355(2)
Chapter 11 Compiler Performance Tuning 357(50)
11.1 Compilers and Optimization
358(2)
11.1.1 Advantages and Disadvantages of Optimization
359(1)
11.2 Optimization Levels
360(7)
11.2.1 Level 0 Optimization
360(1)
11.2.2 Level 1 Optimization (+01)
361(1)
11.2.3 Level 2 Optimization (-0 or +02)
361(4)
11.2.4 Level 3 Optimization (+03)
365(1)
11.2.5 Level 4 Optimization (+04)
366(1)
11.3 Compiling for a Target Runtime Environment
367(2)
11.3.1 +DA
367(1)
11.3.2 +DD
368(1)
11.3.3 +DS
368(1)
11.3.4 +DO
369(1)
11.4 Finer Control Over Optimization
369(20)
11.4.1 Storage Optimizations
370(1)
11.4.2 Standards-Related Options
371(2)
11.4.3 Alias Options
373(3)
11.4.4 Memory Latency Optimizations
376(1)
11.4.5 Symbol Binding Options
377(4)
11.4.6 Overhead Reduction Optimizations
381(2)
11.4.7 Threaded Application Optimizations
383(2)
11.4.8 Floating Point Options
385(2)
11.4.9 Trade-offs Between Memory Expansion and CPU Utilization
387(1)
11.4.10 Bundled Convenience Options
388(1)
11.4.11 Informative Options
388(1)
11.5 Linker Optimizations
389(1)
11.5.1 Options When Linking Shared Libraries
389(1)
11.5.2 Options for Setting the Maximum Virtual Page Size
389(1)
11.6 Profile-based Optimization
390(7)
11.6.1 PBO Issues
392(1)
11.6.2 Changing the Default flow.data File
393(1)
11.6.3 PA-RISC and Itanium Differences When Using PBO
393(1)
11.6.4 Using PBO With Shared Libraries
394(1)
11.6.5 Speeding Up PA-RISC PBO Compilations
394(1)
11.6.6 Execution Frequency Compiler Extensions
395(2)
11.7 Specific Options for Fortran and COBOL
397(1)
11.7.1 Fortran Optimizing Preprocessor
397(1)
11.7.2 Special COBOL Considerations
397(1)
11.8 Why Does Optimization "Break" Applications?
397(2)
11.8.1 C and ANSI C Assumptions
398(1)
11.8.2 Fortran Assumptions
399(1)
11.9 Debugging Optimization Problems
399(1)
11.10 Porting Applications
400(1)
11.10.1 Global and Static Variable Optimization
402(1)
11.10.2 Use of Volatile
402(1)
11.11 Code to Demonstrate Optimization Effects
403(4)
Chapter 12 Java Run-time Performance Tuning 407(10)
12.1 Overview
407(2)
12.2 Performance Issues
409(2)
12.2.1 Garbage Collection Issues
409(1)
12.2.2 Synchronized Keyword
410(1)
12.2.3 Thread Management
411(1)
12.3 Versions on HP-UX
411(1)
12.4 HP-UX Tuneable Parameters and Java
412(1)
12.5 Using Large Java Heap Sizes
412(2)
12.6 Performance Options
414(3)
Chapter 13 Designing Applications for Performance 417(18)
13.1 Tips for Application Design
418(1)
13.2 Shared Versus Archive Libraries
419(1)
13.2.1 Shared Libraries
419(1)
13.3 Archive Libraries
420(1)
13.4 Choosing an Inter-Process Communication (IPC) Mechanism
421(7)
13.4.1 Pipes
421(1)
13.4.2 Message Queues
422(1)
13.4.3 System V Semaphores
422(1)
13.4.4 Memory-Mapped Semaphores
423(1)
13.4.5 Shared Memory
423(4)
13.4.6 Berkeley Sockets
427(1)
13.5 Trade-offs with Network System Calls
428(3)
13.5.1 Select and Poll System Calls
428(1)
13.5.2 Event Port Driver
428(2)
13.5.3 Configuring Berkeley Sockets for Performance
430(1)
13.5.4 Sendfile System Call
431(1)
13.6 Instrumenting an Application for Performance Monitoring
431(4)
13.6.1 Metrics Available by Using the ARM API
431(1)
13.6.2 Run-time libraries
432(1)
13.6.3 Calls Available With the ARM API
432(1)
13.6.4 Using the ARM API to Measure Application Performance
433(2)
Chapter 14 Application Profiling 435(34)
14.1 Caliper
436(18)
14.1.1 Analyzing a Program With Caliper
436(6)
14.1.2 l Using Caliper to Track Down the Performance Issues
442(5)
14.1.3 Attaching to a Running Process
447(1)
14.1.4 Profiling Multiple Processes
447(2)
14.1.5 Making Your Own Configuration Files
449(1)
14.1.6 Instrumenting Sections of an Application With Caliper
449(3)
14.1.7 Detailed Information on Caliper Configuration Files
452(1)
14.1.8 Caliper and PBO
452(1)
14.1.9 Caliper Output Options
453(1)
14.1.10 Advantages and Disadvantages of Caliper
453(1)
14.2 CXperf
454(1)
14.2.1 Steps in Using CXperf
14.2.2 Metrics in CXperf
459(1)
14.2.3 Advantages and Disadvantages of CXperf
459(1)
14.3 gprof
460(1)
14.3.1 Step 1: Special Compilation and Linking
460(1)
14.3.2 Step 2: Program Execution
462(1)
14.3.3 Step 3: Profile Generation
463(1)
14.4 Puma
464(1)
14.5 HPjmeter
465(4)
Appendix A Performance Tools Alphabetical Reference 469(34)
Appendix B HP-UX Version Naming Reference 503(2)
Appendix C Dynamically Tuneable Parameters 505(6)
Index 511

Excerpts

PrefaceMaximizing the performance of Unix systems is a challenge that requires a specialized understanding of operating system behavior, knowledge of the design and tuning capabilities of the applications that run on the system, and an intimate acquaintance with system and application performance measuring tools that assist in tuning. Operating system behavior changes as improvements are made for scalability and significant new features are added. New hardware designs offer the potential for improved performance when the operating system and application takes advantage of the new designs. The behavior of applications can change not only with new versions of the application itself, but also with new versions of the underlying operating system.Although application design and tuning is crucial to getting the best possible performance, the number of applications and application vendors are so numerous that it is impractical to discuss specific application tuning in detail. Therefore, this book focuses on Hewlett-Packard's HP-UX Operating System and the HP 9000 and Integrity Systems on which it runs. This edition adds significant content on networking, Java, new hardware designs, especially Itanium Processor Family (IPF) designs, and HP-UX 11i-specific performance issues. The authors are most familiar with HP-UX, and therefore, the specific operating system architecture and tuning suggestions presented in this book apply directly to HP-UX. The authors also have experience working with a variety of applications and provide guidance on tuning based on their knowledge of how applications and the system interact.We start by presenting a general discussion of performance management in chapters 1 and 2 and by developing a performance management methodology in chapter 3. It is important to realize that performance management is more than crisis-oriented problem solving. Next, in chapter 4, we look at how the operating system kernel is instrumented to provide data that are used to analyze what is happening on the system. Chapter 5 provides suggestions and guidelines for choosing among the many system performance tools that are available and a discussion of the advantages and disadvantages of each tool.Hardware system design is covered in detail in Chapter 6, from component modules through the system itself. Chapters 7 through 10 concentrate on providing an understanding of the major operating system resources (CPU, Memory, Disk and Network) that are utilized by applications and can become bottlenecks to good performance. The authors feel that it is not possible to properly tune a system without understanding how the operating system works. Therefore, operating system design from a performance perspective is discussed in each of these chapters followed by the applicable performance metrics and suggestions for tuning to improve resource utilization.Tuning the operating system alone can provide only a piece of the solution to performance optimization of a system. Application optimization is discussed in the chapters on Compiler Performance Tuning, Java Run-time Performance Tuning, Designing Applications for Performance and Application Profiling. The appendices include an alphabetical reference of the various performance tools, a lexicon of the naming of the versions of HP-UX, and lists of the tuneable parameters that can be dynamically or automatically tuned while the system and its applications are running.We sincerely hope that you find this book to be useful and valuable in keeping your HP-UX systems running optimally.

An electronic version of this book is available through VitalSource.

This book is viewable on PC, Mac, iPhone, iPad, iPod Touch, and most smartphones.

By purchasing, you will be able to view this book online, as well as download it, for the chosen number of days.

Digital License

You are licensing a digital product for a set duration. Durations are set forth in the product description, with "Lifetime" typically meaning five (5) years of online access and permanent download to a supported device. All licenses are non-transferable.

More details can be found here.

A downloadable version of this book is available through the eCampus Reader or compatible Adobe readers.

Applications are available on iOS, Android, PC, Mac, and Windows Mobile platforms.

Please view the compatibility matrix prior to purchase.