PES: A prefetch evaluation system for VM snapshot deduplication

Abstract

Cloud computing has become an increasingly viable and attractive technology for businesses during the past five years. Infrastructure as a Service (IaaS) is one of the important and largely provided cloud services. Most of the IaaS cloud services are based on the virtualization technology, which uses VM images to deploy the Virtual Machines (VM). The state of any VM at a particular time is preserved as a VM snapshot. Since these VM snapshots are larger in size, huge amount of data is being stored in the cloud while backing up the VMs. Hence, an optimization technology, namely, deduplication is essential for the storage of the VM snapshots to remove redundant data among them. Though this technique optimally utilizes the storage space, it incurs huge metadata overhead. Hence, search space of this metadata need to be reduced in order to improve the performance of VM snapshot deduplication. Existing mechanisms, namely, K-Means clustering and similarity based indexing are intended to prefetch the metadata from the disk to the RAM in order to reduce the search space. The objective of this paper is to study and investigate the performance and metadata overhead associated with these two mechanisms in order to suggest the best suited mechanism for VM snapshot deduplication in IaaS cloud.