On Sun, May 16, 2010 at 01:39:30PM +0200, Sander van Vugt wrote:
> On Sun, 2010-05-16 at 11:14 +0000, Andy Smith wrote:
> > On Sat, May 15, 2010 at 11:23:19PM -0400, justino garcia wrote:
> > > Which is faster: iSCSI to linux box over 1Gb or local SATA storage with
> > > intel ICH9?
> > You'll have to benchmark for your load, but local storage is usually
> > faster than remote at least because of the lower latency.
> A simple test that I like doing to measure performance data is like:
> time dd if=/dev/zero of=/mountpoint/dummyfile bs=1M count=1024
> That writes a 1 GB file from the /dev/zero and tells you how long it
> took. I'd be interested to see the differences you measure between iSCSI
> and local storage.
As a very very basic test that something isn't horribly wrong, fair
enough. I would really suggest something like bonnie++ for useful
benchmarks though. The problem with the above is that obviously it
only measures sequential single thread throughput. Increased latency
of requests is a big problem of remote/shared/virtualised storage
which the above won't really expose.
> Of course, I am aware that this is just one simple and limited test, but
> it gives an idea anyway.
bonnie++ is packaged and simple to run so there's really no excuse.
http://bitfolk.com/ -- No-nonsense VPS hosting
"Xandros's low-level support for the Eee mostly seemed to consist of a pile of
shell scripts made of cheese and failure." -- Matthew Garrett
ubuntu-server mailing list
More info: https://wiki.ubuntu.com/ServerTeam