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

FileSystems, Partition sizes, LARGE files??



Hello all,
I've added 4 WD 250Gig IDE 8MB cache hard drives to an existing IDE
fileserver running an ABIT KT7-RAID mainboard with an AMD 1.4GHz CPU and 1
Gig of ram. There is currently 1 existing 120Gig hard drive containing the
main system and current data.

These drives will only be used for archiving large (500MB to 700MB) files
of which there will be at least a few hundred. The system has the ability
to have an additional 3 hard drives added at a later date if needed.

Once setup there will be a second identical server built acting as a
rsynced mirror to be run nightly. This 2nd server is directly connected
via 2nd network cards.

Access to these files is via Samba on a small but fast local network. I'm
currently looking at using the ext3 filesystem dividing each drive into 5
50Gig partitions, setting the Block Size to 4096 and the reserved blocks
percentage option to either 1% or 0%.


1. Is the ext3 filesystem the best choice for this type of job? I've read
where many people have chosen to use the Reiser filesystem for this but I
thought it best suited for small files.

2. I have not used it before but was thinking of installing LVM (Logical
Volume Manager) along with the ext3 filesystem. Comments?

3. Is there a problem with setting the reserved blocks percentage 0% or
should I set it to 1%?

4. Where I'm only using large files, is the Block Size set to 4k by
default ok or should it be changed higher or lower to 2k or 8k?

5. Should the number of inodes be modified in anyway as to not waste
space? I saw this mentioned on another list but I dont quite understand
the benefits.

6 Regarding the size of the files would it be better to create the smaller
50Gig partitions on each drive or used as full 250Gig partitions for each?

Any comments, opinions and suggestions are more than welcome as I'm really
not to sure what the best way to have such a system properly setup.

Many thanks!!!!



Reply to: