EXT3 and Reiser filesystems do journaling. It basically keeps a journal
of files and directories that are accessed so that when the system
crashes, recovery is in less than a second because only the parts that are
used are checked. That means--no more fsck, and fast recovery. It also
is faster, because the locations of the most commonly used files are
cached, and the system "learns" as you use it what stuff you are more
likely to access. This is a medium-sized step in the direction of
large-scale systems with 100% reliability and fault tolerance. SGI has
had it for years, and it has been working very well. The problem that
they are having is that some parts of the code are proprietary(outsourced
to other companies, so SGI does not own the code itself), so they
are only releasing the stuff that they own, with oper-source developers to
do fill in the holes.
Michael J Smith
Flyfisher, Russian Translator, and Linux-Geek-at-Large
2250 Patterson #25
Eugene, OR 97405
(541) 346-7562
