Messages in this thread | | | Date | 01 Jan 1998 16:59:00 +0200 | From | (Kai Henningsen) | Subject | Re: Filesystem optimization.. |
| |
michael@metal.iinet.net.au (Michael O'Reilly) wrote on 30.12.97 in <x7sorby4eu.fsf@metal.iinet.net.au>:
> Richard Gooch <rgooch@atnf.csiro.au> writes: > > > Michael O'Reilly writes: > > > ebiederm+eric@npwt.net (Eric W. Biederman) writes: > > > > MR> Even in this, there's still a win from not needing to allocate a > > > > MR> fixed amount of inodes. > > > > > > > > And again see btree based filesystems. There is reiserfs in the > > > > works, as well as my own shmfs filesystem (though because it has > > > > different prioirties, it doesn't yet keep all inodes in the btree) but > > > > basically with such a beast it is possible, to keep inodes in the > > > > directory tree. > > > > > > I've had a number of people point these out, but there's not a > > > terribly good option for me. I need a stable filesystem, so the > > > smallest possible change for the largest gain. > > > > Your proposed changes to ext2fs would not exactly be "smallest > > possible change". It could introduce all kinds of bugs. > > Yup. But it's a good deal smaller than a brand new filesystem. :)
Actually, I'd say it _is_ a brand new filesystem. It's sufficiently dissimilar to ext2 that I won't count on being able to carry over any assumptions at all, let alone about stability.
MfG Kai
| |