ext2 vs ext3 404
A point. However, decent nightly maintenance (tmpreaper) keepstmp's contents small: # Time files intmp are kept in days. TMPTIME=7 # tmpreaper.conf # - local configuration for tmpreaper's daily run # # This is only used ifetc-cron.daily-tmpreaper was also...
On Thu, 19 Jan 2006 17:21:46 -0500, (PeteCresswell)
The performance difference between EXT2 and EXT3 is minimal, there is no reason to sacrifice the benefits of a journaling file system because of some imaginary performance difference. Unless you are running a fileserver the performance of the file system won't make any difference to you. The other day I tried an experiment where I ran my Verilog regression suite on three different parbreastion types, EXT3, JFS and ReiserFS. The regression takes about 2 1-4 hours and writes about 600MBytes. EXT3 took 2 hours and 15 minutes, JFS took 2 hours and 16 minutes and Reiser took 2 hours and 25 minutes. However EXT3 and JFS were on an SATA-II drive and Reiser was on a PATA drive so I expect all of the performance difference between Reiser and the other two was due to the fact that the PATA drive is a little slower than the SATA drive. The bottom line was that the file system didn't make any noticeable difference.
Desktop systems don't do very much disk I-O and most of what they do is reads. The journal file is only written during writes so you would see 0 difference between EXT2 and EXT3 on read operations. Writes are done in the background so unless your system is running at 100% capacity you won't notice any difference between EXT2 and EXT3 either.
This doesn't seem logical. During what write operations would otherwise-absent steps required to (first) write to a journal not impose overhead? Of course, in pointing out the fallacy posted (a blanket claim...
Linux groups from Newsgroups