|Will you use a separate partition for DB or HTML?|
| 1:06 pm on Jan 14, 2007 (gmt 0)|
If using ext3, will you use a separate partition for
1. DB files
3. Files Caches
| 9:05 pm on Jan 14, 2007 (gmt 0)|
I like to have the OS and applications on a different partition than my own code and data. As an example, I have an 80 gig drive with a 10 gig root partition (basically just the OS), 28 gig /home (mail, mysql data, web files, and user work), and 36 gig /opt (for FTP storage space). I think this is pretty optimal.
I don't see a need to separate HTML, database, and temporary space, unless the database is huge.
| 2:29 pm on Jan 15, 2007 (gmt 0)|
maybe for flexiblity or backup?
or even performance? e.g. use XFS for small but large ammoung of files i.e. cache files?
| 5:08 am on Jan 20, 2007 (gmt 0)|
Here are the seperate filesystems I use as a starting point on ALL Linux systems:
(swap) 1x physical RAM (more if < 1GB physical RAM)
/home Whatever is left
I do think it might make sense to make a seperate partition for database files, if you are running your database on the same machine as your webserver.
The main reason for seperate filesystems is to make the system more resilient in the case where something goes crazy and a filesystem fills up. This is most likely to happen in /tmp or /var. It's also useful to keep /home seperate from / for the same reason.
I put these all in an extended partition (so no restriction on number of filesystems), leaving (potentially) a primary partition for a DOS partition to hold any disk copying, diagnostic, etc. utilities. With most modern Linuxes, you don't need a primary partition at all.
I don't use LVM. Makes it difficult/impossible to use DOS-based tools. Although you lose the ability to dynamically resize and extend filesystems, you can still do so using Partition Magic or other similar tools (I use Paragon). The problem with LVM is that you can't easily manage or back-up filesystems using DOS tools. But most of the DOS-based tools DO support EX2/3 that aren't under LVM.
| 6:30 am on Jan 20, 2007 (gmt 0)|
My partition scheme is much like jtara's, although I prefer to have /usr also in a separate partition.
I use this partition scheme for two reasons. One is speed. Every partition can be formated with the most effective file system and block size. Furthermore on heavily used partitions fragmentations can be easily cured by moving all contents to a temporary disc or partition, reformat the partition and move the contents back.
Also with disc crashes it is handy to have the /, /boot, /etc, /sbin and /bin directories still working when other partitions have died. This makes system recovery much easier.