Business Hypervisors for Real-time Applications

摘要

System virtualization is one of the hottest trends in information technology today. It is not just another nice to use technology but has become fundamental across the business world. It is successfully used with many business application classes where cloud computing is the most visual one. Recently, it started to be used for soft Real-Time (RT) applications such as IP telephony, media servers, audio and video streaming servers, automotive and communication systems in general. Running these applications on a traditional system (Hardware + Operating System) guarantee their Quality of Service (QoS); virtualizing them means inserting a new layer between the hardware and the (virtual) Operating System (OS), and thus adding extra overhead. Although these applications’ areas do not always demand hard time guarantees, they require the underlying virtualization layer supports low latency and provide adequate computational resources for completion within a reasonable or predictable timeframe. These aspects are intimately intertwined with the logic of the hypervisor scheduler. In this paper, a series of tests are conducted on three hypervisors (VMware ESXi, Hyper-V server and Xen) to provide a benchmark of the latencies added to the applications running on top of them. These tests are conducted for different scenarios (use cases) to take into consideration all the parameters and configurations of the hypervisors’ schedulers. Finally, this benchmark can be used as a reference for choosing the best hypervisor-application combination.