[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[DotGNU]VRS DFS: paper of interest (Google file system)
From: |
Eric Altendorf |
Subject: |
[DotGNU]VRS DFS: paper of interest (Google file system) |
Date: |
Tue, 30 Sep 2003 08:43:45 -0700 |
User-agent: |
KMail/1.4.3 |
Hi folks, still lurking, sorry for the lack of participation. I saw
this and thought it might be of interest to anyone still thinking
about the design of the distributed file system for VRS.
Good points in relation to the VRS: (1) it's distributed over
unreliable hardware and connections, (2) it makes use of the
underlying filesystem and stores there large "chunks" (64MB) of the
very large (multi-gig) "actual" files (which, for the VRS, could
actually be essentially encrypted disk or filesystem images).
Bad points: it's designed primarily for streaming -- appending to the
end of large files, and reading long continuous streams of data from
a file. Small files, middle-of-the-file writes, and caching of data
is avoided.
---------- Forwarded Message ----------
O | S | D | N NEWSLETTER September 30, 2003
The Design Of The Google File System
http://slashdot.org/article.pl?sid=03/09/29/2227224
[0]Freddles writes "This is [1]an interesting paper (PDF)
describing the design approach to [2]Google's file system. The
design had to take account of requirements for huge file sizes, a
highly responsive infrastructure and an assumption that hardware
components will always fail."
Links
0. http://www.fredhoysted.com
1. http://www.cs.rochester.edu/sosp2003/papers/p125-ghemawat.pdf
2. http://www.google.com/
--
Eric Altendorf // http://www.speedtoys.com/~eric
- [DotGNU]VRS DFS: paper of interest (Google file system),
Eric Altendorf <=