it is not so much the miniscule space that will be recovered, it is the
latency involved with doing operations on this directory.
create a temporary directory, and within it, create a few hundred thousand
tiny files. profile a little program that runs stat() or otherwise in
here, then delete a majority of the files. run your profiled program
again.
now run your profiled prg on a new directory with only a dozen files in
it.
on servers such as mail, news, squids, etc., such file operations latency
can have a very measurable impact.
consider IRL, a hand sized deck of 3x5 index cards. it's quick and easy
to sort through them and find what you are looking for. now get a box
that holds several reams of paper and fill it with those cards. even if
99% of those cards have had their information erased, it still takes
significantly longer to retrieve the desired information.
caching helps but cached filespace can be nulled and it has to be built
again.
with the rapidly growing availability of data to manage, we need to be
proactive in design, not reactive. when ext3+btree finally appears for
development, how much data are we going to be managing? i'll wager it's
far more than it is now. people have an ever growing penchant for
information...look at the US government amongst others.
-d
--
Please read the linux/Documentation/ files and review the last week of mail
on linux-kernel before posting your problem.
PLEASE don't quote _many_ lines and type _few_ lines
-thx
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html