Boston Linux & Unix (BLU) Home | Calendar | Mail Lists | List Archives | Desktop SIG | Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings
Linux Cafe | Meeting Notes | Blog | Linux Links | Bling | About BLU

BLU Discuss list archive


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Linux file systems



 On Thu, 31 Jul 2008 08:24:30 -0400 (EDT) 
[hidden email] wrote: 

> I am using Ubuntu now, and have used a number of distros in the past. I've 
> played with file systems and a few years ago I did some benchmarks for a 
> project. 
> 
> At that time: 
> ReiserFS was good for many small file, but performed badly for larger 
> files. Also a high write and file creation environment performed badly. 
> ReiserFS was a bit buggy. 
> 
> JFS and XFS behaved similarly to each other, in that big files and 
> moderately large amounts of moderately large/small files worked well. 
> Worked well in a high write and file creation environment. IBM's JFS 
> seemed more stable and with a better tool chain. 
> 
> EXT3 was a stodgy all around lame performer. Was one of the worst 
> performers in dynamic environments. 
> 
> EXT2 was had pretty good performance but that can be attributed to a lack 
> of journaling. 
> 
> For discussion, what is the general consensus on file systems now? Are the 
> above assumptions still valid? Opinions? 
> 
> I have a project that may require a million plus directories. Ideally, I'd 
> like to have them all at the same level and perform well, but if I have 
> too I can use the hierarchical hash-bucked strategy, i.e. top/0/0/0, 
> top/0/0/1, top/0/0/2 ... top/0/1/0, etc. to keep the number of files per 
> directory less than 1000. 


BLU is a member of BostonUserGroups
BLU is a member of BostonUserGroups
We also thank MIT for the use of their facilities.

Valid HTML 4.01! Valid CSS!



Boston Linux & Unix / webmaster@blu.org