Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

Background: We have gigs of data, which require to process large files simultaneously and in parallel with similar processes. We are going to get SSD just because of fastest latency.

We will install CentOS 5 over it.

Question: What is the best File System available for CentOS 5, that will enhance our processes?

share|improve this question
    
Isn't this just another way of saying serverfault.com/questions/5958/… –  romandas Feb 6 '10 at 1:53
add comment

4 Answers

up vote 5 down vote accepted

ssd has much larger 'physical data allocation unit' - so regardless of the fileystem you choose remember to align your partitions with size of sector on your ssd drive. if you don't - your write performance will suffer a lot.

this is even true for newest desktop disks with 4kB sector instead of 512B sector - just yesterday i did empirical experiment that gave me 29MB/s write speed with default alignment [ from 63rd logical 512B sector ] vs 70MB/s with proper alignment [ starting from 128rd sector ].

regarding the file system - ext4 is supposed to support trim, but i never really tested it. take a look here and here.

share|improve this answer
    
The link thunk.org/tytso/blog/2009/02/20/… is dead. –  N.N. Jun 30 '11 at 20:48
    
@N.N. then check web.archive.org : web.archive.org/web/20100112035708/http://thunk.org/tytso/blog/… –  pQd Oct 27 '11 at 21:38
add comment

I believe JFS or XFS are known to have the best performance when operating on large files. In addition to the selection of an appropriate filesystem for your task, I'd recommend also throwing as much RAM as possible into the server as you can, even if you don't need it for your processing tasks. Linux will use this extra RAM for disk cache, which will improve your performance.

share|improve this answer
add comment

Use an SSD as you would any hard drive. It doesn't matter. Nor does the filesystem you use with the you choose. The questions you need to actually answer are: What is the nature of the data I'll be storing/reading/writing? Is it a bunch of small files? One huge continuous large file?

Then figure out (from the questions already asked on SF) which filesystem has the best performance for the type of data you're processing.

share|improve this answer
add comment

The important thing to consider is whether your disk controller will perform wear-leveling and error correction. If it does, a solid-state file system is less of a concern. If you do require a special file system, log-stuctured FSs are generally recommended. Such file systems include JFFS2, NILFS, and YAFFS. Also, check out the tool bonnie++ to test drive performance with different filesystems.

ref:

http://en.wikipedia.org/wiki/Flash_file_system <-general

http://www.linux-mag.com/cache/7345/1.html <-nilfs

share|improve this answer
2  
Wear leveling is really a non-issue. All SSDs have wear leveling built into their internal controllers. The hard disk controller doesn't need to know anything about wear leveling. –  EEAA Feb 6 '10 at 2:17
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.