A Transparent Remote Paging Model for Virtual Machines

In virtual machine systems, with the increase in the number of VMs and the demands of the applications, the main memory is becoming a bottleneck for the application performance. To improve paging performance and alleviating thrashing behavior for memory-intensive or I/O-intensive virtual machine workloads, we proposed hypervisor based remote paging, which allows a virtual machine to transparently use the memory resource on other physical machines as a cache between its virtual memory and virtual disk device. The goal of remote paging is to reduce disk accesses, which is much slower than transferring memory pages over modern interconnect networks. As a result, the impact of thrashing behavior can be alleviated since the average disk I/O latency is reduced. Remote paging also benefits some I/O intensive applications, resulting in a better-than-native performance. Our remote paging model is totally transparent to the guest operating system as well as its applications, and is compatible with existing techniques like ballooning, hypervisor disk cache and page sharing. A combination of them can provide a more flexible resource management policy.