Performance Analysis for Javaв„ў Websites

Note: Code listings are numbered in a separate series in each chapter and gathered in their own subsection, following the figures.

Figure 1.1 Customers in a traditional bookstore

Figure 1.2 Customers using an on-line bookstore

Figure 1.3 Concurrent users in a traditional bookstore

Figure 1.4 Concurrent web site load

Figure 1.5 Active customers in a traditional bookstore

Figure 1.6 Active web site clients

Figure 1.7 Traditional store customers per hour

Figure 1.8 Throughput dynamics without waiting

Figure 1.9 Throughput dynamics with waiting

Figure 1.10 Throughput curve: brick and mortar store

Figure 1.11 Typical web site throughput curve

Figure 1.12 Total checkout time in a traditional store

Figure 1.13 Response time in an on-line store

Figure 1.14 Think time example

Figure 1.15 Response time queue for a brick and mortar store

Figure 1.16 Throughput curve and response time graph

Figure 1.17 Steady-state measurement interval

Figure 1.18 Reducing the checkout process path length

Figure 1.19 Example bottleneck: gift-wrap department

Figure 1.20 Traditional store handling more customers in parallel

Figure 1.21 Scaling with multiple bookstores

Figure 1.22 Horizontal scaling example

Figure 2.1 Model-View-Controller within a Java web application

Figure 2.2 Small servlets versus front controller strategy

Figure 2.3 Threading/queuing "funnel"

Figure 2.4 A frozen web site

Figure 2.5 Servlet loading versus JSP loading

Figure 2.6 Cookies and HTTP sessions

Figure 2.7 Failover with a persistent HTTP session database

Figure 2.8 Enterprise JavaBean transactions without a fa §ade Bean

Figure 2.9 Enterprise JavaBean transactions using a fa §ade Bean

Figure 2.10 Clones in both vertical and horizontal scaling

Figure 3.1 Router placement in a large web site

Figure 3.2 A typical web site firewall configuration

Figure 3.3 Reverse proxy server used for security

Figure 3.4 Caching reverse proxy servers

Figure 3.5 Different speeds for networks handling different loads

Figure 3.6 A load balancer distributing HTTP traffic

Figure 3.7 An example of load balancer affinity routing

Figure 3.8 Proxy server impacting even load distribution with affinity routing

Figure 3.9 Users accessing a web site through a proxy server farm

Figure 3.10 A simplified version of the SSL handshake

Figure 3.11 Plug-in operation

Figure 3.12 Vertical scaling of the application server

Figure 3.13 A small web site using horizontal scaling

Figure 3.14 A DMZ configuration with servlets, JSPs, and EJBs

Figure 4.1 Two examples of web site traffic patterns

Figure 4.2 A conceptualized JVM heap and corresponding settings

Figure 4.3 Typical garbage collection cycles

Figure 4.4 Garbage collecting too frequently

Figure 4.5 Typical memory leak pattern

Figure 5.1 A brokerage web site's traffic patterns

Figure 5.2 Yearly traffic patterns for an e-Commerce site

Figure 5.3 Web site configuration supporting access by pervasive devices

Figure 6.1 Peak versus average web site load

Figure 6.2 One HTML assembled from multiple HTTP requests

Figure 6.3 The same daily traffic volume spread over both 8- and 24-hour days

Figure 6.4 An example test environment

Figure 7.1 Java Pet Store demo home page

Figure 7.2 Java Pet Store hierarchy

Figure 7.3 Example of primitive scripts

Figure 7.4 LoadRunner parameter properties GUI

Figure 7.5 Result of Browse for Fish

Figure 7.6 Example of customizing think times

Figure 8.1 Script recording

Figure 8.2 Example LoadRunner ramp-up

Figure 8.3 Example of SilkPerformer V workload configuration

Figure 8.4 Testing environment

Figure 8.5 Sample LoadRunner Summary Report

Figure 8.6 Sample LoadRunner monitor

Figure 9.1 An example test environment

Figure 9.2 Shared network impact

Figure 9.3 Typical DMZ configuration for application servers

Figure 9.4 Single-disk versus multidisk database I/O

Figure 9.5 Testing live legacy systems

Figure 9.6 Master/slave load generator configuration

Figure 9.7 Testing the networks with FTP

Figure 9.8 One possible test ordering for a large test environment

Figure 11.1 Iterative test and tuning process

Figure 11.2 Test run measurement interval

Figure 11.3 Measuring over long runs to simulate steady state

Figure 11.4 Example of high run-to-run variation

Figure 11.5 Example of acceptable run-to-run variation

Figure 11.6 Example of run variation with a downward trend

Figure 11.7 Results after introducing a change

Figure 11.8 Throughput curve to determine saturation point

Figure 11.9 SMP scaling comparisons

Figure 11.10 Vertical scaling using multiple JVMs

Figure 11.11 Load balancing architecture options

Figure 11.12 Testing individual server performance

Figure 11.13 Quantifying load balancer overhead

Figure 11.14 Testing a two-server cluster

Figure 11.15 Two-server cluster scalability results

Figure 12.1 Sample vmstat output

Figure 12.2 Sample Microsoft Windows 2000 System Monitor

Figure 12.3 Sample JVM heap monitor

Figure 12.4 Sample Thread Analyzer screen capture

Figure 12.5 Sample Resource Analyzer output

Figure 13.1 Underutilization example

Figure 13.2 Example of bursty behavior

Figure 13.3 Example of a bursty back-end database system

Figure 13.4 Example of high CPU utilization

Figure 13.5 Example of high system CPU utilization

Figure 13.6 Example of high CPU wait on a database server

Figure 13.7 Uneven cluster loading

Figure 13.8 IP-based versus user -based affinity

Figure 14.1 TriMont performance test configuration

Figure 15.1 Sample user ramp-up performance test results

Figure 15.2 User ramp-up performance test results

Figure 15.3 TriMont scalability results

Figure D.1 Summary test results graph