Performance Detective: Practical steps to avoid system bottlenecks

Here are some steps you can take to help you quickly identify system bottlenecks and find out where to find help when your system is blocking.

When the http://www.aliyun.com/zixun/aggregation/32593.html "> system runs too slowly, you must determine which of the many components in the configuration may be the culprit. The disk I/O may be too slow or the system may be low on memory. is a poorly written application created? Or is it a runaway process? There appears to be an adapter firmware fix that resolves all problems.

Bottlenecks may be one of these issues, or they may be a combination or some other problem. When performance problems start to rage, users may be under a lot of pressure to fix it. If you use a forward-looking approach to tuning system performance, you can help you share this pressure. Developing a vision plan and early intervention can help you avoid huge performance bottlenecks. Step back. It is also valuable to consider whether the system expects to fit the workload.

Adopt A forward-looking Approach

Here are some practices to help you prevent performance problems, at least to help you respond quickly when performance problems arise.

Record configuration

It is important to share your knowledge with people who can understand and manage your configuration, and the best way is to record your configuration, paying special attention to areas that use non-default settings or do not comply with industry standards.

The Ibm®power systems™ Server provides many tools to help you record configurations. For example, you can use commands, such as prtconf, to list the configuration of a virtual machine, as shown in Listing 1.

Listing 1. prtconf Command Display System configuration

# Prtconfsystem Model:ibm,9117-mmamachine Serial number:01a02b3processor type:powerpc_power6processor Implementation Mode:power 6Processor Version:pv_6_compatnumber of processors:2processor Clock speed:4208 mhzcpu type:64-bitkernel Type : 64-bitlpar info:4 a3_everest-nimmemory size:8192 mbgood Memory size:8192 mbplatform modularity level:em350_108firmware version:ibm,em350_108

Documents should include disk configuration, environment variables, operating system tunable settings, device properties, application versions, system models, types, and serial numbers. If your system is performing poorly, you may need to ask for technical support, which will help you get the details of the system with little effort.

Using the Change Management program

With appropriate change management, you can alert other stakeholders who are about to make changes, provide step-by-step details of the change, and prepare a workable rollback plan.

Implement recommended tuning settings

By implementing the recommended tuning parameters, you can better leverage existing hardware and ease resource pressures before you start building.

Even with a high-performance storage subsystem, I/O bottlenecks may be encountered at the operating system level, and a common example is when I/O queues for a physical volume are filled.

The Iostat command in Listing 2 shows that the service queue for this physical volume cannot respond to service requests it receives. See Sqfull value in the last column of the queue value. This means that other I/O requests must wait.

Listing 2. Iostat shows the full queue

Hdisk89 xfer:%tm_act bps tps bread bwrtn 75.0 43.4M 336.5 6.1K 43.4M read:rps avgserv minserv maxserv Timeouts fails 1.5 123.9 15 .1 1.0S 0 0 write:wps avgserv minserv maxserv timeouts fails 335.0 28.1 1.5 275.0 0 0 queue:avgtime mintime maxtime avgwqsz Avgsqs Z sqfull 43.9 0.0 297.1 2.5 1.6 239.0

You can use the Chdev command to increase the queue depth, as shown in Listing 3.

Listing 3. Increase Queue Depth

Chdev-l hdisk89-a queue_depth=20