Template-Based Memory Deduplication Method for Inter-Data Center Live Migration of Virtual Machines

摘要: Live migration of virtual machines (VMs) can benefit data centers through load balancing, fault tolerance, energy saving, etc. Although live between geographically distributed enable optimized scheduling resources in a large area, it remains expensive and difficult to implement. One the main challenges is transferring memory state over WAN. There conflict low transmission speed WAN rapid change contents. This paper proposes novel method with page-count-based deduplication, which takes advantage fact that VMs running same or similar operating systems other software tend have identical pages. Template pages are selected based on number occurrences each page across multiple indexed by content hash. When transferred, source host first compares templates. If match identified, transfers index instead page. The experimental results show our approach reduces time 27% transferred 38% average compared default QEMU-KVM.