Ok, my results so far, here’s the output for a run on a 2.0.6 site straight from git, i.e. the out of the box single home page site, this has been run from another machine on my local lan on the server described above :-
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking 192.168.0.20 (be patient)
Server Software: Apache/2.2.17
Server Hostname: 192.168.0.20
Server Port: 80
Document Path: /performance
Document Length: 318 bytes
Concurrency Level: 5
Time taken for tests: 10.980 seconds
Complete requests: 3000
Failed requests: 0
Write errors: 0
Non-2xx responses: 3000
Total transferred: 1650000 bytes
HTML transferred: 954000 bytes
Requests per second: 273.23 [#/sec] (mean)
Time per request: 18.299 [ms] (mean)
Time per request: 3.660 [ms] (mean, across all concurrent requests)
Transfer rate: 146.76 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 4 9 1.8 9 28
Processing: 4 9 1.8 9 29
Waiting: 4 9 1.8 9 29
Total: 10 18 2.2 18 36
Percentage of the requests served within a certain time (ms)
50% 18
66% 19
75% 19
80% 20
90% 21
95% 22
98% 24
99% 25
100% 36 (longest request)
As can be seen, no crashes. I’ve tried this on the server itself both localhost and its external IP, on a revo site containing clarity pack and on an evolution site, all from the same server, a different machine on my local lan and externally, basically quite a few combinations. The timings vary obviously(on localhost the test takes 0.287 seconds for instance) but I’ve not seen any crashes of apache, neither do I see any appreciable differences between a revo site with clarity pack and an evo site with its demo content, certainly not what’s been observed above. I’ve also tried different sites I have installed that are not MODx related.
Ok, It looks as though for whatever reason revo is as happy on this machine as anything else is running this particular benchmark.
It looks as though it comes back to the server and its environment, not revo per see.
I think we need to note all this and run the same test on different servers and server environments, only then will we be able to come to more definite conclusions as to what exactly in the server environment revo is susceptible to, if anything.