On Sun, 28 Feb 1999, Ingo Molnar wrote:
# > I believe that ext2 is a decent filesystem; however, it handles directory
# > searching using a linear algorithm. Each search for a name through that
# > 15000 element directory will be executed as a naive linear search.
#
# fortunately, the way how ext2fs accesses inodes is a secondary issue,
# Linux 2.2 uses a dynamic namecache to access inodes. Thus if you system
# has enough memory to cache 15000 dentries, it will perform very fast.
Just to add information to the pool: I created a 30,000 entry directory
and did some manips on it. Here's the results:
** creation
root(5)@neko [144 /<2>tmp/ll] # /bin/time ../lots
2.71user 371.50system 6:15.25elapsed 99%CPU
** listing
root(5)@neko [235 /<2>tmp/ll] # /bin/time ls -1 | wc -l
0.65user 0.14system 0:00.78elapsed 100%CPU
30000
** removal (had to use find as rm * produced too many arguments)
root(5)@neko [250 /<2>tmp/ll] # /bin/time find . -type f -exec rm {} \;
63.38user 87.38system 2:29.43elapsed 100%CPU
This is on a fairly normal PII 233 system with 192MB ram, running on an
IDE drive (WDC36400L) under 2.2.2ac5.
G'day!
-- n i c h o l a s j l e o n
elegance through simplicity*http://mrnick.binary9.net*[EMAIL PROTECTED]
good fortune through truth*roaming:[EMAIL PROTECTED]*ICQ#2170994*U+($++)
TRA#6805*not all questions have answers*pseudogeek:P+++($++)L+($++)W=lm@b9
trust no-one with an iq under 150*understand yourself before trying others
-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to [EMAIL PROTECTED]
