The goal of this thesis, is to research how the performance of different types of workloads can be influenced by different qemu configurations and, based on the results if this research, provide a UI for automatic optimization of a virtual machine.

Student will:

• study how different qemu configurations (such as NUMA/vNUMA configuration, IO Threads, emulator thread pinning, virtual HW presence etc) is supposed to affect the performance
• apply this understanding to perform series of tests researching how this configuraion actually affects performance of different types of workloads (database, compute intensive, application server) with a goal to find the most performant configuration
• formulate profiles with suggested configuration for different types of workloads
• implenment a UI providing the end user with a possibility to optimize the selected VM to a specific type of workload