Stanford University Department of Electrical Engineering
Computer Systems Colloquium
4:15PM, Wednesday, October 26, 2005
NEC Auditorium, Gates Computer Science Building B03
http://ee380.stanford.edu
The Reiser4 Filesystem
Ways In Which Extra Rigor In Scientific Methodology Can Consume Years Of Your Life,
And How The Result Can Be So Very Worthwhile
If you want a filesystem to handle names that are semi-structured data queries, then you need to have a storage layer that can handle both large and small files. Traditional filesystems handle small files very poorly, and traditional databases handle large files poorly.
Reiser4 fixes the performance issues with traditional balanced tree algorithms that made them perform poorly for filesystem usage patterns, and then implements a filesystem on top of that.
The result is a filesystem with roughly twice the performance of other Linux filesystems. Future work will be to add semi-structured data queries into the filesystem.
About the speaker:
Hans Reiser entered UC Berkeley after completing the eighth grade and received a BA in Systematizing, which was an individual major. His senior thesis discussed differences in the philosophy of computer science and the physical sciences as illustrated by name space (file system) design issues.
He founded Namesys and hired a small team of programmers to implement the Reiserfs file system for Linux. Benchmarks are at www.namesys.com/benchmarks.html. The Resierfs file system is based on balanced tree algorithms rather than block allocation. Hans is primary architect, project manager, and one of the programmers for the file system. Prior to founding Namesys, Hans held positions at Synopsis, IBM Research, Premos, ARDC, and the UCB Open Computing Facility. Hans is a practicing Interactive Media artist and has participated in a number of exhibitions and shows. |
Contact information:
Hans Reiser
reiser@namesys.com
from Hacker News https://ift.tt/8EUWKlw
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.