<< Determining the Cause of Performance Problems | Chapter 10 - Designing Performance Testing Suites >>
<< Determining the Cause of Performance Problems | Chapter 10 - Designing Performance Testing Suites >>
Monitoring Computer Resources and Tuning Your System
milliseconds (ms). This does not necessarily mean that the system response is the
same. Further evaluation of the results reveals that System A has response times of 11,
12, 13, and 12, and System B has response times of 1, 20, 25, and 2. Although the mean
time is the same (12 ms), the minimum, maximum, and standard deviation are all
quite different.
rather than software design. For example, your disk job service times could be
unacceptably slow due to a concentration of disk transfers being sent to a single disk
rather than being spread across several disk drives. This problem is typically fixed by
relocating some of the frequently accessed files (such as swap files or temporary files)
to a disk with less activity.
resulting in substantial network congestion. Even the simplest round-trip delay from
client to server and back can take several seconds. This problem is typically fixed by
splitting an overloaded LAN segment into two or three segments with routers in
between. Sometimes you need to add a second network card to server systems so they
can be directly accessible to two LAN segments without going through a router.
that cannot be fixed by changing the software design.
response time data. You can monitor your computer resource use during a suite
playback and then graph this data over the corresponding virtual tester response
times, to determine whether imbalance in the hardware resources is causing slow
response times.
Computer Resource Usage onto Response Time on page 342.