服务器过多如何监控性能

liangshi

请问是Linux服务器还是Windows服务器？
微软推出了System Center Operation Manager （SCOM）可以完美的解决服务器（支持Linux和Windows）性能监控问题，但是价格不菲。
另一种途径是利用Windows自带的perfmon来监控。或者，自Windows 2008之后的操作系统，有一个命令typeperf。它的帮助如下。从帮助可以看出，它支持收集其他机器上的performance counter，而且机器名、counter名可以从配置文件中读取。你可以预先写好几个配置文件，不同的测试让typeperf加载不同的配置文件，这样就可以监控不同的机器和counter。typeperf的输出可以是csv格式的文本文件，适合进一步分析。

Usage:
typeperf { <counter [counter ...]> | -cf <filename> | -q [object] | -qx [object] } [options]

Parameters:
  <counter [counter ...]>       Performance counters to monitor.

Options:
  -?                            Displays context sensitive help.
  -f <CSV|TSV|BIN|SQL>          Output file format. Default is CSV.
  -cf <filename>                File containing performance counters to monitor, one per line.
  -si <[[hh:]mm:]ss>            Time between samples. Default is 1 second.
  -o <filename>                 Path of output file or SQL database. Default is STDOUT.
  -q [object]                   List installed counters (no instances). To list counters for one
                                object, include the object name, such as Processor.
  -qx [object]                  List installed counters with instances. To list counters for one
                                object, include the object name, such as Processor.
  -sc <samples>                 Number of samples to collect. Default is to sample until CTRL+C.
  -config <filename>            Settings file containing command options.
  -s <computer_name>            Server to monitor if no server is specified in the counter path.
  -y                            Answer yes to all questions without prompting.

Note:
  Counter is the full name of a performance counter in
  "\\<Computer>\<Object>(<Instance>)\<Counter>" format,
  such as "\\Server1\Processor(0)\% User Time".

Examples:
  typeperf "\Processor(_Total)\% Processor Time"
  typeperf -cf counters.txt -si 5 -sc 50 -f TSV -o domain2.tsv
  typeperf -qx PhysicalDisk -o counters.txt