Windows Server 2019: introduction to System Insights

In Windows Server 2019 has been included a new feature called System Insights which it introduces natively in the operating system predictive capabilities. Thanks to an accurate analysis that occurs locally to the system, based on a machine learning model, is able to provide, with a high level of reliability, forecasts of problematic conditions that may occur in the Windows Server environment. This article shows how to enable this feature and the main features of the solution.

Install System Insights

System Insights in Windows Server 2019, does not require specific installation requirements, and can be activated on physical or virtual systems, so the Hypervisor or cloud platform on which reside agnostic. Installation is simple and can be done using one of the following ways:

  • Through Windows Admin Center.

Figure 1 – Enabling System Insights through Windows Admin Center

  • With PowerShell, using the command “Add-WindowsFeature System-Insights-IncludeManagementTools”
  • Adding the feature System-Insights by using Server Manager.

Using System Insights

After installation, you can proceed by configuring the desired settings for forecasts of the CPU capacity, the use of networking and storage consumption.

Figure 2 – Weather forecast available

System Insights is able to provide the results of the analyses carried out and their predictions after some hours of activation.

The possible states that can be assumed by all forecast are as follows:

Figure 3 – Possible States

For each estimate available you can select the scheduling of when it performed:

Figure 4 – Prediction scheduling

In addition, you can configure scripts that are executed when returning a specific status code, useful to take corrective actions automatically.

Figure 5 – Actions to be taken against certain States

Each type of capacity can also be invoked manually in a forced manner via the button Invoke.

By selecting the different available forecasts, you can view the detailed information.

Figure 6 – CPU usage example

Figure 7 - Sample of Critical status for the consumption of space on the volume E:

Figure 8 – Details that bring the prediction on the exhaustion of space in the next 7 days

System Insights is able to provide this predictive information based on machine-learning models, analyzing different elements such as performance counters and events. All data are collected and analyzed locally to the machine, without iterations with elements in the cloud and with a non-significant resource consumption. Using PowerShell, you have the ability to aggregate the results of System Insights from multiple virtual machines. In that regard, it shows a sample script to aggregate the results of multiple systems.

Figure 9 – Sample PowerShell script for aggregating data from System Insights

This approach can be useful for a small number of systems, but if you want to have an overview of this information for more complex environments, you can bring together the information of System Insights in a workspace of Azure Log Analytics. To do this, simply set up your Log Analytics workspace to also collect events generated by System Insight (Microsoft-Windows-System-Insights/Admin):

Figure 10 – Configuration of the workspace of Log Analytics

In this way you can easily generate the Rule to be notified based on specific queries.

Figure 11 – System Insights event ID

Figure 12 -Example of a query that is used in a Rule

Conclusions

For system administrators, the instrument System Insight is useful and easy to use, enabling you to predict several problematic conditions that may occur on your Windows Server, all in a fully integrated manner in the operating system. With this feature you can achieve greater continuity of service and a reduction of the time required to clear error conditions.