On 09/26/2012 05:28 PM, John Mark Walker wrote:
On Wed, Sep 26, 2012 at 2:23 PM, Berend de Boer <[email protected] <mailto:[email protected]>> wrote: >>>>> "John" == John Mark Walker <[email protected] <mailto:[email protected]>> writes: John> This is different from any other benchmark I've seen. I John> haven't seen that much of a disparity before. What benchmarks? Steve's experience is very similar to what everyone sees when trying out gluster. Have you seen write:read ratios > 5:1? I certainly haven't. I have seen discrepancies, sure, but not by that much.
I've seen stuff like this. Looks like a caching issue (gluster client) among other things.
Read performance with the gluster client isn't that good, write performance (effectively write caching at the brick layer) is pretty good.
I know its a generalization, but this is basically what we see. In the best case scenario, we can tune it pretty hard to get within 50% of native speed. But it takes lots of work to get it to that point, as well as an application which streams large IO. Small IO is a (still) bad on the system IMO.
I've not explored the 3.3.x caching behavior (largely turned it off in 3.2.x and previous due to bugs which impacted behavior).
-- Joseph Landman, Ph.D Founder and CEO Scalable Informatics Inc. email: [email protected] web : http://scalableinformatics.com http://scalableinformatics.com/sicluster phone: +1 734 786 8423 x121 fax : +1 866 888 3112 cell : +1 734 612 4615 _______________________________________________ Gluster-users mailing list [email protected] http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
