On Mon, Nov 17, 2003 at 06:00:09AM -0500, [EMAIL PROTECTED] wrote:
> Hi Colm...
>
> Slainte!...
> Cead mile failte romhat!
> Go raibh maith agat!
Agus t� f�in a ch�irde, chaitfidh m� r� b'�idir gurb seo on
t-aon deis a bh�is gam cumars�id le Gaeilgeoir so comh-th�acs
seo, ach mar a dearta� �fach - Is annamh is �ontach.
Actually I must get round to tranlating the errordocs ;)
> Wow... I believe everything you are saying... and
> please don't take this the wrong way... but I'm not
> sure a test that only runs for 1.1 second and 1000
> requests with 100 clients being launched ( on the
> same machine? ) is a good way to get accurate results
> especially in the TPS ( Transactions Per Second )
> numbers. The rounding errors alone could be huge
> with so little time on the clock.
Saves me time though ;) It's a quick way to get numbers - and I
was trying to be fairer to 1.3, but with 100,000 -
Apache 2 prefork:
Concurrency Level: 100
Time taken for tests: 100.192136 seconds
Complete requests: 100000
Failed requests: 0
Write errors: 0
Total transferred: 190466640 bytes
HTML transferred: 145650960 bytes
Requests per second: 998.08 [#/sec] (mean)
Time per request: 100.192 [ms] (mean)
Time per request: 1.002 [ms] (mean, across all concurrent
requests)
Transfer rate: 1856.45 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 44 12.8 47 134
Processing: 22 54 15.7 51 305
Waiting: 2 49 15.2 47 304
Total: 29 99 6.6 98 314
Percentage of the requests served within a certain time (ms)
50% 98
66% 99
75% 100
80% 100
90% 104
95% 110
98% 117
99% 126
100% 314 (longest request)
Apache 1.3: same hardware
First try:
lasaire:/usr/local/apache# ../apache2/bin/ab -c 100 -n 100000 http://samhain:8080/
This is ApacheBench, Version 2.0.40-dev <$Revision: 1.131 $> apache-2.0
Copyright (c) 1996 Adam Twiss, Zeus Technology Ltd,
http://www.zeustech.net/
Copyright (c) 1998-2002 The Apache Software Foundation,
http://www.apache.org/
Benchmarking samhain (be patient)
apr_poll: The timeout specified has expired (70007)
Total of 4060 requests completed
Second attempt:
Concurrency Level: 100
Time taken for tests: 188.725417 seconds
Complete requests: 100000
Failed requests: 0
Write errors: 0
Total transferred: 187087890 bytes
HTML transferred: 145668432 bytes
Requests per second: 529.87 [#/sec] (mean)
Time per request: 188.725 [ms] (mean)
Time per request: 1.887 [ms] (mean, across all concurrent
requests)
Transfer rate: 968.09 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 87 23.0 90 178
Processing: 4 99 25.3 97 751
Waiting: 1 93 24.0 91 690
Total: 10 187 8.9 186 847
Percentage of the requests served within a certain time (ms)
50% 186
66% 188
75% 189
80% 190
90% 194
95% 201
98% 206
99% 209
100% 847 (longest request)
And this benchmark is light compared to what we see as production
load. I could run the benchmarks into the many millions, but it's
not going to make a difference.
> Rasmus recent benchmark shows the EXACT
> OPPOSITE and I think you have certainly just proved
> that something is seriously wrong with THAT test...
> but I'm not sure yours is the end-all be-all proof either.
They're different tests, it's not uncommon for different tests
to have different results. The atomics around dynamic content
are very different from static, Apache 2.0 kicks 1.3 ass when
it comes to static content. When I migrate dynamic content to
2.0 I don't see the same improvment, but I do see some. Then
again, I'm a heavy user of suexec stype dymanic content rather
than modules.
--
Colm MacC�rthaigh Public Key: [EMAIL PROTECTED]
