On Fri, Jun 12, 2009 at 07:05:10PM -0700, Adam Williamson wrote: > On Fri, 2009-06-12 at 17:25 -0400, Paul W. Frields wrote: > > On Fri, Jun 12, 2009 at 07:55:56PM +1000, Eric Springer wrote: > > > I agree with the sentiment that phoronix reviews are lazy, poor, etc > > > -- but that doesn't mean nothing is revealed by them. Especially > > > considering how many people will use these benchmarks to make > > > conclusions about Fedora, we should make sure it presents as best as > > > it can. So I think it is important to establish why our apache > > > result were so poor and what can be done to fix it. > > > > "Our" Apache results on the Phoronix tests, AIUI, are from an Apache > > they compiled, which is not what most people are going to use. > > There's also no mention of whether they mitigated the way results > > would have changed from our use of SELinux. Poorly stated, what I meant was not "mitigated" but "addressed". In some cases the raw benchmarks may not tell the whole story, such as the tradeoff for audit that Adam mentions, but... > However, back a few posts, someone tested with the Fedora packaged > apache and reproduced the results - same result as Phoronix got, it was > slow. The current thread thinking is that audit is the cause of this. ...regardless, the same results happen with our packaged httpd means we should at least try to identify the culprit. Mainly, I would like in general to see more context around benchmarking before it's touted as authoritative. I'm speaking from the general perspective of someone not enamored of sound-bite news. -- Paul W. Frields http://paul.frields.org/ gpg fingerprint: 3DA6 A0AC 6D58 FEC4 0233 5906 ACDB C937 BD11 3717 http://redhat.com/ - - - - http://pfrields.fedorapeople.org/ irc.freenode.net: stickster @ #fedora-docs, #fedora-devel, #fredlug -- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
