Metadata Efficiency in Versioning File Systems

  • Authors:

  • Craig A. N. Soules;Garth R. Goodson;John D. Strunk;Gregory R. Ganger

  • Affiliations:

  • Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University

  • Venue:
  • FAST '03 Proceedings of the 2nd USENIX Conference on File and Storage Technologies
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

Versioning file systems retain earlier versions of modified files, allowing recovery from user mistakes or system corruption. Unfortunately, conventional versioning systems do not efficiently record large numbers of versions. In particular, versioned metadata can consume as much space as versioned data. This paper examines two space-efficient metadata structures for versioning file systems and describes their integration into the Comprehensive Versioning File System (CVFS), which keeps all versions of all files. Journal-based metadata encodes each metadata version into a single journal entry; CVFS uses this structure for inodes and indirect blocks, reducing the associated space requirements by 80%. Multiversion b-trees extend each entrys key with a timestamp and keep current and historical entries in a single tree; CVFS uses this structure for directories, reducing the associated space requirements by 99%. Similar space reductions are predicted via trace analysis for other versioning strategies (e.g., on-close versioning). Experiments with CVFS verify that its current-version performance is sim-ilar to that of non-versioning file systems while reducing overall space needed for history data by a factor of two. Although access to historical versions is slower than con-ventional versioning systems, checkpointing is shown to mitigate and bound this effect.