Path: utzoo!utgpu!water!watmath!clyde!rutgers!rochester!cornell!batcomputer!pyramid!voder!cullsj!gupta
From: gupta@cullsj.UUCP (Yogesh Gupta)
Newsgroups: comp.os.misc
Subject: Re: Contiguous files; extent based file systems
Summary: Random access to data in large files
Message-ID: <177@cullsj.UUCP>
Date: 17 Dec 87 22:02:25 GMT
References: <561@amethyst.ma.arizona.edu> <3228@tut.cis.ohio-state.edu> <9828@mimsy.UUCP>
Distribution: na
Organization: Cullinet Software, San Jose, CA
Lines: 14

In article <9828@mimsy.UUCP>, chris@mimsy.UUCP (Chris Torek) writes:
> Personally, I have never cared whether my files were contiguous.  All
> I care about is that they be reasonably fast to access.
> 

I definitely agree.  However, I find that if I create a 100MB file
under Unix (BSD 4.2, System V Rel 1), the overhead in
randomly accessing various parts of it is too high (due to indirect
inode structures).  Any comments?  What if I know that my file will
be of the order of 100MB and then declare an extent size of 1MB (wastage
of 0.5%, on the average) in an extent based file system?
-- 
Yogesh Gupta			| If you think my company will let me
Cullinet Software, Inc.		| speak for them, you must be joking.