In this chapter, we are going to cover the following recipes:
Managing performance has been a challenge since the earliest days of computing. Windows NT 3.1 came with a marvelous tool, Performance Monitor (Perfmon), that allowed you to see what the OS and applications are doing and what resources they are consuming.
This chapter shows you how you can use PowerShell to obtain and display performance information. The recipes in this chapter show you how you can use a cmdlet and WMI to get and report the performance information. This chapter shows how you can generate performance graphs and reports for management. And of course, you can leverage the various recipes in this chapter to conduct your performance monitoring and reporting.
The Windows performance monitoring framework is known as Performance Logging and Alerting (PLA). PLA is built into Windows and uses COM and DCOM to obtain performance and diagnosis information from both local and remote computers.
PLA enables you to obtain a variety of performance data from running systems. PLA...
Get-Counter is the cmdlet you use both to discover the counter sets available on a machine, and to obtain performance samples from a local or remote server. In this recipe, you use a Windows Server 2016 server, SRV1, to examine performance counter sets and counters on local and remote computers.
This recipe uses several remote machines: DC1, CA, SRV1, FS1, FS2, and PSRV. Adjust the recipe to reflect the computers in your testing or production environment.
This recipe uses several servers to simulate a normal organization. Consider using fewer servers.
Get-Counter to discover performance counter sets on the local machine:$CounterSets = Get-Counter -ListSet * "There are {0} counter sets on [{1}]" ` -f $CounterSets.count, (hostname)
$Machines = 'DC1', 'CAa', 'SRV1', 'FS1', 'FS2', 'PSRV' foreach ($Machine in $Machines...
Another way to access performance information is via WMI. You can use either the WMI or the CIM cmdlets to access a large number of performance counters, as an alternative to using Get-Counter. The naming structure is different from using Get-Counter. With WMI, each counter is a separate WMI class.
With WMI, each performance counter set is a WMI class. The WMI performance counters are found in the ROOT\CimV2 namespace and have a name that begins with Win32_Perf. For example, the Memory performance counter set contains 36 separate counters. The WMI class Win32_PerfFormattedData_PerfOS_Memory contains 46 properties including all of the individual performance counters. With WMI, you get all the measurements back in one call to Get-CimInstance, whereas you would need to call Get-Counter for each sample. There are other ways to collect counters as shown in later recipes.
In this recipe, you get performance counters from local and remote machines using...
The first two recipes in this chapter used different techniques (Get-Counter and WMI) to retrieve specific counters and counter sets. As you seen, getting a large number of counter values for detailed analysis can be very slow with these mechanisms. These techniques are ideal for retrieving one or two bits of information (CPU utilization for example, or pages/second). If you want to get a larger number of statistics (for example, all of the networking statistics including TCP, UDP, IP, and ICMP) the techniques do not scale well.
A better approach to gathering large number of counters is to use the Data Collector Sets and have Windows do the work for you. To do this, you first create and configure a collector set. When you start the set, Windows starts collecting the data you have configured the collector set to return. Finally, when the collection has been completed, you use Performance Monitor to view the results.
This approach is very much easier...
In the first three recipes in this chapter, you have seen different ways to collect performance information. In this recipe, you create a report on this gathered performance information based on the CSV files output by the data collection process.
This recipe was written to use SRV1. As with any recipe, you can use any server in your infrastructure, but you would need to adjust the details of all steps to reflect the changes you are making. This recipe also used CSV files, as created by the PLA infrastructure. As noted earlier, it is simple and straightforward to adjust the counter output file to be CSV.
Use the recipe Configuring and using Data Collector Sets, change the counter output file to CSV and generate CSV output in the folder C:\PerfLogs\ADMIN.
In the previous recipe, you created a simple text based report which you could expand to cover not just the CPU on the SRV1 server, but more counters across multiple machines. But they would be pure text. You could use performance monitor and the binary log files to create Perfmon graphs you could cut/paste into a report.
In this recipe, you use the data generated using the data collector mechanism to draw a graph using classes from the Windows.Forms.DataVisualization namespace.
Like the Reporting on performance data recipe, this recipe uses CSV files from the data collection process noted earlier.
System.Windows.Forms and System.Windows.Forms.DataVisulization assemblies:Add-Type -AssemblyName System.Windows.FormsAdd-Type -AssemblyName System.Windows.Forms.DataVisualization
0:$CSVFile = Get-childitem -Path C:\PerfLogs\Admin\*.csv `
-Recurse...The PLA subsystem that you have been working with in this chapter has an additional kind of report that the PLA and PowerShell can generate, a System Diagnostic report. This report is simple to create and makes use of some of the approaches used in this chapter.
$PerfReportName="System\System Diagnostics" $DataSet = New-Object -ComObject Pla.DataCollectorSet$DataSet.Query($PerfReportName,$null)$DataSet.Start($true)
Start-Sleep -Seconds $Dataset.Duration$Dataset.Query($PerfReportName,$null)$PerfReport = $Dataset.LatestOutputLocation + "\Report.html"& $PerfReport
© 2017 Packt Publishing Limited All Rights Reserved
