Exposing file system mappings with MapFS

Authors:
Jake Wires;Mark Spear;Andrew Warfield
Affiliations:
University of British Columbia;University of British Columbia;University of British Columbia
Venue:
HotStorage'11 Proceedings of the 3rd USENIX conference on Hot topics in storage and file systems
Year:
2011

Citing 5
Cited 0

Application performance and flexibility on exokernel systems

Proceedings of the sixteenth ACM symposium on Operating systems principles
Zero copy I: user-mode perspective

Linux Journal
Xen and the art of virtualization

SOSP '03 Proceedings of the nineteenth ACM symposium on Operating systems principles
Redundancy elimination within large collections of files

ATEC '04 Proceedings of the annual conference on USENIX Annual Technical Conference
Generating realistic impressions for file-system benchmarking

FAST '09 Proccedings of the 7th conference on File and storage technologies

Quantified Score

Hi-index	0.00

Visualization

Abstract

The conventional model of a file as a contiguous array of bytes hides information about the physical location of data from users. While this simplifying abstraction can be useful in some cases, it can also lead to suboptimal performance and unnecessary overhead. A growing number of applications - even those as basic as the Unix cp utility - can benefit from increased access to file system metadata. We present MapFS, a file system which allows applications to create, inspect, modify, and remove the mappings established between individual files and physical storage. MapFS gives users increased power and flexibility, facilitates true end-to-end application design, and optimizes many common file system tasks.