How is 'Load Average' calculated by Java Machine Agent installed on Windows server?

[Saradhi.Potharaju]

Finding the load average calculated by a Java Agent on a Windows server requires the following data:

1. ProcessorQueueLength counts how many threads are ready in the processor queue, but not currently able to use the processor.
   • To find this value, launch PowerShell and run the following command:
     

     Get-WmiObject -Query "Select * from Win32_PerfRawData_PerfOS_System" | findstr ProcessorQueueLength​

2. The value of mathematical constant e= 2.718281828459045.
   
   
3. Load average intervals are:
   • 1 minute  (60 sec)
   • 5 minutes (300 sec)
   • 15 minutes (900 sec)
     
     
4. "Sample frequency" is the frequency with which the samples are collected for each interval.
   
   Example: In 1 minute, 2 samples were collected. In 5 minutes, 10 samples were collected. Sample frequency is 30 seconds (default).
   
   This can be set in the <machine_agent_home>/extensions/ServerMonitoring/conf/ServerMonitoring.yml file by changing the "sampling interval" property.
   
   
5. Load average is set at 0 before computing. Then, for each sample collected at the sampling frequency, the load average is calculated as an exponentially-damped moving average. The data points are added into the load average results for each sample.
   

   exp = e power -(sampleFrequency/interval)
   
   loadAverage = loadAverage * exp
   loadAverage = loadAverage + ProcessorQueueLength*(1-exp)​