Implementing GPUs in your virtual desktop environment (for example vGPU or GPU pass-though) allows better use of graphical intensive applications (like Catia, AutoCAD, Photoshop, etc.). These applications have an impact on utilization (CPU, memory, disk I/O) in your infrastructure. Not only the applications will impact your infrastructure, sending the graphical intensive data over the LAN or WAN will at least impact your network and the CPU on the end-point and inside your virtual desktop. In this article I will explain the impact on network level.
If you’re used to working with VMware, you probably know about the nifty little utility esxtop. This tool allows you to view real-time performance data on a ESX host. It also allows you to gather performance metrics in batch mode, which allows you to analyze the data in eg. Excel or ESXplot.
For those who are working with XenServer, a similar utility is available called xentop. The xentop utility allows you to monitor real-time performance metrics, like esxtop does. However, gathering data like in esxtop is a challenge.