Home
| Calendar
| Mail Lists
| List Archives
| Desktop SIG
| Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings Linux Cafe | Meeting Notes | Linux Links | Bling | About BLU |
I agree that a partition used for database does not need, and probably should not use a journalling file system if the file system is dedicated to the database. Databases really need (or better would like) raw access to the hardware. it really does not matter from a performance standpoint what filesystem you use for relatively static partitions, such as /boot. Same goes for the root file system is /home and /var are in separate file systems. One issue for laptops is that ext3 has default maximum mount count and a check interval. So, if you boot that laptop twice a day (work and home), you might have a full fsck once a month or more frequent. But, you can change both the interval and max count using tune2fs. IHMO: Most users and servers probably are not going to see much benefit from ReiserFS or JFS in normal operations. In cases where you have specific needs, then certainly the choice of file system and parameters needs to be done. On Thu, 1 May 2008 09:42:31 -0400 (EDT) [hidden email] wrote: > I'm not a big fan of EXT[3|4] except as the boot partition as that is not > very dynamic. I have been using IBM jfs for a long time now and find it > better/faster/etc than EXT3, even on database, web, and mail servers. > > IMHO: > EXT2 is great for a database journal in that you won't be double > journalling. (I often speculate that a very minimal UNIX file system > designed for purely for speed and regularly sized blocks, something like a > streamlined FAT system, would be awesome for databases.) > > EXT3 is good for system boot partitions as it does not need fsck on on a > restart and the volume is likely not very dynamic. > > ReiserFS (last I did any research) was pretty good when you had a lot of > small files. > > IBM JFS, again, last time I did any testing, was a better choice for a > generic file system as it had a pretty good balance of journal speed and > large vs small file storage/access. > > SGI's XFS I sort of abandoned (I admit I was a wimp) because I thought IBM > jfs would have better and more active development.
BLU is a member of BostonUserGroups | |
We also thank MIT for the use of their facilities. |