|
From: | Gordan Bobic |
Subject: | Re: [Gluster-devel] Workloads for GlusterFS |
Date: | Fri, 24 Dec 2010 08:26:29 +0000 |
User-agent: | Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2.4) Gecko/20100628 Red Hat/3.1-1.el6 Lightning/1.0b2 Thunderbird/3.1 |
On 12/24/2010 01:53 AM, Mukund Buddhikot wrote:
Hello Gluster Folks, One can use dd, iometer, "internal" data sets and OLTP workloads for performance testing of GlusterFS. Is there a "standardized" workload/benchmark for DFS such as GlusterFS..
The ones you mention are "standard" benchmarks. They are mostly useless for gauging real-world performance. The only worthwhile testing is testing for your particular application.
But if you really want a "redneck" test, I find kernel-building on a machine with lots of fast cores with a suitable -j parameter works quite well. You will need a machine that has enough CPU to saturate the I/O, though (4+ cores running at 3+ GHz).
Gordan
[Prev in Thread] | Current Thread | [Next in Thread] |