simple performance test
Added by hellocyf about 15 years ago
server
AMD Athlon(tm) 64 Processor 3000+ 1g ram 100mbit
Linux ubuntu 2.6.27-9-generic #1 SMP Thu Nov 20 22:15:32 UTC 2008 x86_64 GNU/Linux
client
AMD Sempron(tm) Processor 3000+ 1g ram 100mbit
Linux ubuntu 2.6.28-4-generic #11-Ubuntu SMP Fri Jan 16 21:50:52 UTC 2009 x86_64 GNU/Linux
config
both config file is modify ./doc/lighttpd.conf
enable mod_status and set server.event-handler = "linux-sysepoll"
addional
in 1.5.0 set server.network-backend = "gthread-aio"
modify mod_status, so the test out put is same like below
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
<title>Status</title>
</head>
<body>
<h1>lighttpd 1.4.21</h1>
<table summary="status" border="1">
<tr>
<th colspan="2">Server-Features</th>
</tr>
</table>
</body>
</html>
http_load -fetches 10000 -parallel 1000 urls
urls is url http://192.168.1.101/server-config
result
1.5.0
10000 fetches, 100 max parallel
cpu 55%
10000 fetches, 100 max parallel, 4.42e+06 bytes, in 1.67401 seconds
442 mean bytes/connection
5973.68 fetches/sec, 2.64037e+06 bytes/sec
msecs/connect: 0.439808 mean, 11.448 max, 0.083 min
msecs/first-response: 15.788 mean, 19.656 max, 0.691 min
HTTP response codes:
code 200 -- 10000
100000 fetches, 100 max parallel
cpu 100%
100000 fetches, 100 max parallel, 4.42e+07 bytes, in 16.8454 seconds
442 mean bytes/connection
5936.33 fetches/sec, 2.62386e+06 bytes/sec
msecs/connect: 0.462623 mean, 56.472 max, 0.088 min
msecs/first-response: 15.9809 mean, 68.881 max, 0.668 min
HTTP response codes:
code 200 -- 100000
10000 fetches, 604 max parallel, 4.42e+06 bytes, in 4.42319 seconds
442 mean bytes/connection
2260.81 fetches/sec, 999278 bytes/sec
msecs/connect: 130.928 mean, 3001.61 max, 0.108 min
msecs/first-response: 24.3267 mean, 3001.61 max, 0.289 min
HTTP response codes:
code 200 -- 10000
100000 fetches, 727 max parallel, 4.42e+07 bytes, in 27.1934 seconds
442 mean bytes/connection
3677.37 fetches/sec, 1.6254e+06 bytes/sec
msecs/connect: 81.2122 mean, 21000.5 max, 0.093 min
msecs/first-response: 34.6875 mean, 12000.4 max, 0.423 min
HTTP response codes:
code 200 -- 100000
1.4.21
10000 fetches, 100 max parallel, 4.43e+06 bytes, in 1.33824 seconds
443 mean bytes/connection
7472.51 fetches/sec, 3.31032e+06 bytes/sec
msecs/connect: 0.535982 mean, 17.93 max, 0.11 min
msecs/first-response: 12.0286 mean, 27.068 max, 0.77 min
HTTP response codes:
code 200 -- 10000
100000 fetches, 100 max parallel, 4.43e+07 bytes, in 13.5882 seconds
443 mean bytes/connection
7359.31 fetches/sec, 3.26017e+06 bytes/sec
msecs/connect: 0.556935 mean, 44.945 max, 0.085 min
msecs/first-response: 11.8775 mean, 3015.28 max, 0.612 min
HTTP response codes:
code 200 -- 100000
10000 fetches, 202 max parallel, 4.43e+06 bytes, in 4.29158 seconds
443 mean bytes/connection
2330.14 fetches/sec, 1.03225e+06 bytes/sec
msecs/connect: 17.0419 mean, 3000.53 max, 0.107 min
msecs/first-response: 15.883 mean, 3001.25 max, 0.294 min
HTTP response codes:
code 200 -- 10000
100000 fetches, 257 max parallel, 4.43e+07 bytes, in 15.8805 seconds
443 mean bytes/connection
6297.04 fetches/sec, 2.78959e+06 bytes/sec
msecs/connect: 10.0183 mean, 8999.25 max, 0.106 min
msecs/first-response: 15.2607 mean, 9026.64 max, 0.312 min
HTTP response codes:
code 200 -- 100000
conclusion
this test shows that 1.4.21 has better performace than 1.5.0
It is strange.