Posted by: Mike | July 30, 2008

Linux Filesystem Desktop Benchmarks

I’ve crafted a set of benchmarks.  I’m using iozone as my synthetic benchmark du jour, honestly I just don’t feel like messing with bonnie++.  Then I’m doing some basic work with the linux kernel source as a use case scenario, measuring copies and moves and such.  Then I’m also adding to the mix a bunch of MP3’s and m4a’s: my collection of music under the artist name of Neil Young.  Nautilus reports “163 items, totalling 876.9 MB” on the directory, and I’ll just trust it because I don’t feel like counting.

I am not using the machine during testing (obviously) because I want the testing to be fair.  However, I am also not creating an irrationally low load on the computer.  I’m running Ubunu Hardy 64bit, with Gnome, compiz, etc.  No screensaver, because that might be unevenly balanced.  But I want a realistic desktop load on a reasonably modern desktop machine.

I got frustrated with the benchmarks I could find on the internet.  They were targetted at mail servers running Dual Xeons and SCSI drives and/or were all a bit out of date.  So a) I’ll have useful data for myself, b) possibly share useful information for others who are looking at filesystems for bulk desktop storage: media files, local source code repos, and such.

I’ve scripted the entire thing, and will share that along with the raw data, and basic analysis.

Oh, and my machine: An Intel DP35DPM motherboard with an Intel Q6600 CPU, 4GB of Crucial DDR2-667 memory, and the testing is being done on a WD7500AAKS 750GB 16MB cache SATAII hard drive.  Not crazy fast, but a modern desktop.

Now off to let this thing run.  Since I’m testing xfs, jfs, ext2, ext3, reiserfs, reiser4, vfat, and ntfs, I wonder how long it will take.  I threw in ext2, vfat, and ntfs just for comparison.  I don’t expect anyone to actually use these. reiser4 was as bleeding edge as I am willing to go.  ext4 and btrfs (which looks promising long term) are a little too bleeding edge, so I’m not bothering with them.  Honestly, who wants to store bulk data on something not yet reliable.  Studying them is purely academic, and this is meant to be practical.  Hopefully I’ll have a writeup with pretty charts and graphs by the end of the week.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s


%d bloggers like this: