MemX: Supporting Large Memory Workloads in Xen Virtual Machines

Modern grid computing and enterprise applications increasingly execute on clusters that rely upon virtual machines (VMs) to partition hardware resources and improve utilization efficiency. These applications tend to have memory and I/O intensive workloads, such as large databases, data mining, scientific workloads, and web services, which can strain the limited I/O and memory resources within a single VM. In this paper, we present our experiences in developing a fully transparent distributed system, called MemX, within the Xen VM environment that coordinates the use of cluster-wide memory resources to support large memory and I/O intensive workloads. Applications using MemX do not require specialized APIs, libraries, recompilation, relinking, or dataset pre-partitioning. We compare and contrast the different design choices in MemX and present preliminary performance evaluation using several resource-intensive benchmarks in both virtualized and non-virtualized Linux. Our evaluations show that large dataset applications and multiple concurrent VMs achieve significant speedups using MemX compared against virtualized local and iSCSI disks. As an added benefit, we also show that live Xen VMs using MemX can migrate seamlessly without disrupting any running applications.