An Experimental Analysis of the Xen and KVM Latencies

The recent developments in virtualisation technologies have made feasible the execution of complex and performance critical applications in virtual machines. When such applications are characterised by real-time constraints, the virtual machines must be scheduled predictably over the physical cores. Several works in real-time literature have analysed such a scenario, proposing advance scheduling and design techniques to respect the application constraints. However, most of the previous works focused on scheduling algorithms and theoretical analysis, without considering important implementation details such as the latencies introduced by the virtualisation mechanism. The paper, which can be seen as a complement for such works, investigates the latencies introduced by two of the most widely used hypervisors, Xen and KVM, so that previous theoretical analysis and algorithms can be used in practice.