Venti: A New Approach to Archival Storage
FAST '02 Proceedings of the Conference on File and Storage Technologies
ATEC '06 Proceedings of the annual conference on USENIX '06 Annual Technical Conference
Opening black boxes: using semantic information to combat virtual machine image sprawl
Proceedings of the fourth ACM SIGPLAN/SIGOPS international conference on Virtual execution environments
Secure on-demand grid computing
Future Generation Computer Systems
The effectiveness of deduplication on virtual machine disk images
SYSTOR '09 Proceedings of SYSTOR 2009: The Israeli Experimental Systems Conference
Managing security of virtual machine images in a cloud environment
Proceedings of the 2009 ACM workshop on Cloud computing security
Multi-layered Virtual Machines for Security Updates in Grid Environments
SEAA '09 Proceedings of the 2009 35th Euromicro Conference on Software Engineering and Advanced Applications
Difference engine: harnessing memory redundancy in virtual machines
Communications of the ACM
Experiences with content addressable storage and virtual disks
WIOV'08 Proceedings of the First conference on I/O virtualization
A study of practical deduplication
FAST'11 Proceedings of the 9th USENIX conference on File and stroage technologies
An empirical analysis of similarity in virtual machine images
Proceedings of the Middleware 2011 Industry Track Workshop
CLOUDCOM '11 Proceedings of the 2011 IEEE Third International Conference on Cloud Computing Technology and Science
Finding collisions in the full SHA-1
CRYPTO'05 Proceedings of the 25th annual international conference on Advances in Cryptology
Hi-index | 0.00 |
Allowing users to build custom virtual machines as execution environments for their tasks provides flexibility for users and providers of Infrastructure-as-a-Service Clouds or virtualized Grid computing environments. On the downside of this flexibility are the increasing storage requirements for virtual machines. This problem is further exacerbated if version histories of virtual machines are kept to facilitate reproducibility of scientific results. Additionally, the simplicity of virtual machine creation provided by corresponding tools invites users to create multiple virtual machines for different purposes, further increasing their numbers. However, the traditional way of storing virtual machines as image files does not scale well with an increasing number of virtual machines. Several approaches have been proposed to solve this problem, each with its own drawbacks. In this paper, the Marvin Image Store (MIS) is presented to efficiently store a large number of Linux virtual machine images including their version history, independent of the distribution and the type of file system. The MIS minimizes the space required to retain images by importing them into its repository using a file based deduplication technique. Layered virtual machine images are used to reduce the time to import (updated) images and to reassemble them from the compositional manifests stored in the MIS. Furthermore, the possibility to directly mount stored images can skip the reassembly process completely. Experimental results indicate that the storage requirements can be reduced by up to 94% compared to the original images. The import of layered virtual machine images is up to 78% faster than the import of regular virtual machine images, and the export is up to 84% faster.