Mercurial: A Traffic-Saving Roll Back System for Virtual Machine Cluster
Bin Shi, Deqing Chen, Lei Cui, Jingsheng Zheng, Bo Li · 2014
Virtual Machine Cluster (VMC) is now widely used to host network applications due to its well scalability and high availability compared to physical cluster. To provide fault tolerance, VMC snapshot is one well known technique, it saves the entire VMC state into stable storage and rollbacks the VM from the latest saved state upon failures. However, due to the large snapshot size as well as numerous VM snapshots, the existing rollback approaches always utilize large amount of network resources, and further incur long rollback latency since network resource is limited. In this paper, we present Mercurial, a rollback system that finds the identical pages across the VM snapshots and then multicast a single copy of the duplicated pages to associated virtual machines, thereby saving the network traffic and reducing the rollback latency. Our approach is effective especially when the same OS and similar applications are running in the VMC. We have implemented Mercurial on QEMU/KVM and conducted several experiments to evaluate its effectiveness and efficiency. The experimental results demonstrate that our approach can save 34% network traffic introduced by transferring the identical pages repeatedly and reduce by 28% of the restore latency.