Unless you have a third-party tool installed on your Windows server, you won’t have any clue it’s about to crash until it does. This isn’t so devastating when you have several servers in a web farm. The crashed server leaves the rotation, and the others take up the slack. However, if you only have one or two servers handling client requests, it only takes one server crash to stop productivity. Before your server crashes, here is how you can perform a quick performance check in Windows.
Reviewing CPU and Memory Usage in Task Manager
Most Windows users are familiar with Task Manager. Task Manager has been a part of Windows for years. It’s the perfect way to close frozen programs to avoid rebooting your machine. It’s also a great way to check your server for any suspicious usage spikes.
Right-click the Windows taskbar and choose “Start Task Manager.” Click the Performance tab. You’ll see a window similar to the image below.
Note that the image above is from a server that doesn’t have very much memory installed. For this reason, memory resources are overloaded. If this was an actual server, the memory usage would be a concern. The server’s applications are using 87% of the physical memory, which doesn’t leave any room for other applications or traffic. In this scenario, you would install more memory to improve server performance.
The top part of the image shows CPU usage. CPU usage is low for this machine. With only 35% CPU usage, the server has enough resources to handle additional applications or traffic from the network. Notice that the usage spikes are consistent. Windows servers have terrible performance when CPU usage spikes to 100%. This commonly happens when you run too many applications on the server. For instance, a database requires much of the server’s resources. If you run other applications in addition to a database, the server can have 100% CPU spikes, which kills performance.
Checking Network Resources in Task Manager
Even with memory and CPU usage at efficient levels, your network card could be overloaded. This happens if someone is attacking the network servers, or you have a rogue network resource that’s configured incorrectly.
Using the same Task Manager window, click the “Networking” tab.
Notice that this server is running with a link speed of 54 Mbps. The network card only has a maximum of 25% usage spikes. There are three network cards installed on this server, which is why there are three sections. However, the server only uses the wireless network card. If your server is using multiple cards, you will see multiple graphs in this window that let you know how each card is functioning.
If your server is attacked by a hacker or a misconfigured network resource sends rogue packets, this graph will show you extreme spikes in the network card resources section. When this happens, the network card drops legitimate packets and your users will complain that they are randomly disconnected from services.
What Can You Do If Resource Spikes Are Too High?
Most administrators check these numbers when the server is already overloaded. The best way to manage Windows servers is to have some kind of third-party software that checks resources continuously and sends you an alert when they are overloaded. If you don’t have access to third-party software, the administrator should perform a quick check each morning to ensure that the servers are running smoothly.
Once you determine that your server is overloaded, you need to know if more hardware is the answer or an application is at fault. In rare cases, the hardware could be faulty and needs replacement. Unfortunately, you won’t know it’s faulty hardware causing the issue until you actually replace it.
Before you start adding more CPU, memory and network card resources, you can perform a few checks to identify if software is the problem. Before you make any of these changes, take the server offline. Some of these changes will stop services to your clients.
If your server is a web host, check your web services. You can reboot the application pool in Internet Information Services (IIS).
You can right-click any of the application pools in the list and recycle them. This action essentially reboots the web service. Just remember that recycling an application pool takes it offline for the amount of time it takes for the service to reboot. If you have a critical application running on your server, only perform this step during off-peak hours or after you’ve told users that the application will be down for several minutes. It actually takes a few seconds to recycle an application pool, but telling users it will take a few minutes gives you some extra time.
Task Manager has a tab named “Processes” that also gives you clues to what could be causing performance issues. If a process is taking too much server resources, you can close it and see if your server’s performance returns. Go back to your Task Manager window and click the “Processes” tab.
In the window above, only Chrome is taking memory resources. There is nothing that jumps out as overloading the server. However, if you see an application that takes too much memory, you can right-click it and choose “End Process.” This stops the application and process from running on your server. When you stop a process, the application no longer runs. Ensure that the process you stop is not a critical one for your users. You can later restart the application if your users need it.
If none of these solutions work, the last effort should be to reboot the server. You should only reboot a server if it’s absolutely necessary, but sometimes a reboot “fixes” the issues.
Quick performance checks on your server can save you from dealing with a total crash in the future. These checks give you insight to what could be causing resource usage spikes, so you can stop them before they become a critical outage on your network. Use these quick steps to review your Windows server and take the necessary steps to stop performance leaks.