Data management in a distributed file system for storage area networks
Data management in a distributed file system for storage area networks
SOSP '03 Proceedings of the nineteenth ACM symposium on Operating systems principles
The Panasas ActiveScale Storage Cluster: Delivering Scalable High Bandwidth Storage
Proceedings of the 2004 ACM/IEEE conference on Supercomputing
OASIS: implementation of a cluster file system using object-based storage devices
ICCSA'06 Proceedings of the 6th international conference on Computational Science and Its Applications - Volume Part I
IEEE Communications Magazine
| Hi-index | 0.00 |
Using the standard Object-based Storage Device, OASIS has been developed as a cluster filesystem. Like the most of existing out-of-band cluster filesystems using ODSs, the OASIS could not support the extended remove UNIX semantics to defer the remove of an inode until the uses of the inode in all client nodes are finished. This nonsupport generates the problems that it does not protect users to make use of the deleted inode and does share an inode of a deleted directory entry with a newly created entry, which is due to client node's VFS to support the remove UNIX semantics. To resolve these problems, this paper proposes the re-designed OASIS to perform an inode deletion until its uses are finished by extending the existing lock table for cache coherence. The suggested approach can support the remove UNIX semantics in the distributed environment and easily be adopted in the existing out-of-band cluster filesystems if using their locking mechanism.