Sascha Ottolski wrote:
i did a simple test recently, which suggests that there is a significant
performance difference: I did a comparison of client vs. server
side afr with bonnie, for a one client and two servers setup with tla
patch628, connected over GB Ethernet; please see my results below.
There also was a posting on this list with a lot of test results,
suggesting that server side afr is fastest:
http://lists.nongnu.org/archive/html/gluster-devel/2007-08/msg00136.html
Yeah, I wouldn't rely on that post for much.
1) It was a long time ago (WRT GlusterFS development), a lot has changed.
2) It was done using VERY simplistic volume specifications (No
performance xlators at all).
3) It doesn't simulate real world usage well at all, the test program
was a very simple perl script made up for for that purpose.
Something your bonnie tests didn't include was a baseline for a server.
You didn't specify what type of throughput we should expect from the
disks, so we have no way of knowing whether glusterfs is being
constrained by the server's disks. It looks like it's utilizing close
to 80% of the theoretical maximum for GB ethernet in the case of the
client side AFR though, and consideration of the path the data follows
and the maximum throughput of that medium is another thing to consider
when deciding whether to put AFR on the server or the client.
--
-Kevan Benson
-A-1 Networks
