Hace tiempo que quería testear el Lite Speed. Había oído hablar muy bien de el. Se trata de un servidor web con licencia que trae una interfaz de administración y opciones interesantes como compilar php al vuelo mediante la interfaz de una forma muy simple.
Pero lo que me interesaba era el rendimiento. Y me he llevado alguna que otra sorpresa. El software de pruebas es un WordPress funcionando bajo PHP 5.3.3 y APC 3.1.9. En dos servidores Centos 6 con características parejas.
La base de datos es MysQL 5.5 en un caso y 5.1 en otro aunque comparten el mismo my.cnf. Por otro lado la instalación de Lite Speed es por defecto sin tunear.
Ambos WordPress fueron testeados; en el caso del contenido dinámico; sin caché de por medio, es decir, sin WP CACHE, W3 TOTAL CACHE, o similares. Para probar el rendimiento sin cachés de por medio (aunque sigamos teniendo el APC y el QUERY CACHE, hice las pruebas después de un reinicio de servidor web)
1. Contenido dinámico
# ab -c10 -n10 http://rubenortiz.es/
This is ApacheBench, Version 2.0.40-dev <$Revision: 1.146 $> apache-2.0
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Copyright 2006 The Apache Software Foundation, http://www.apache.org/
Benchmarking rubenortiz.es (be patient).....done
Server Software: Apache/2.2.3
Server Hostname: rubenortiz.es
Server Port: 80
Document Path: /
Document Length: 0 bytes
Concurrency Level: 10
Time taken for tests: 7.70842 seconds
Complete requests: 10
Failed requests: 0
Write errors: 0
Non-2xx responses: 10
Total transferred: 5330 bytes
HTML transferred: 0 bytes
Requests per second: 1.41 [#/sec] (mean)
Time per request: 7070.841 [ms] (mean)
Time per request: 707.084 [ms] (mean, across all concurrent requests)
Transfer rate: 0.71 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 0 0.0 0 0
Processing: 3107 5141 1232.0 5369 7070
Waiting: 3104 5140 1232.5 5368 7069
Total: 3107 5141 1232.0 5369 7070
Percentage of the requests served within a certain time (ms)
50% 5369
66% 5516
75% 6261
80% 6265
90% 7070
95% 7070
98% 7070
99% 7070
100% 7070 (longest request)
# ab -c 10 -n 10 https://www.rubenortiz.es/
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking www.rubenortiz.es (be patient).....done
Server Software: LiteSpeed
Server Hostname: www.rubenortiz.es
Server Port: 80
Document Path: /
Document Length: 83936 bytes
Concurrency Level: 10
Time taken for tests: 8.564 seconds
Complete requests: 10
Failed requests: 0
Write errors: 0
Total transferred: 843600 bytes
HTML transferred: 839360 bytes
Requests per second: 1.17 [#/sec] (mean)
Time per request: 8564.082 [ms] (mean)
Time per request: 856.408 [ms] (mean, across all concurrent requests)
Transfer rate: 96.20 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 1 2 0.6 2 3
Processing: 5191 6880 1763.4 8550 8561
Waiting: 5002 6765 1837.5 8496 8545
Total: 5192 6883 1763.6 8552 8564
Percentage of the requests served within a certain time (ms)
50% 8552
66% 8553
75% 8554
80% 8555
90% 8564
95% 8564
98% 8564
99% 8564
100% 8564 (longest request)
1.1 Resumen test
Apache 2.2.3 => 1.41
LiteSpeed => 1.17
OO! Pues la verdad no me esperaba yo esto. Apache funciona un poco mejor que Lite Speed en contenido dinámico sin cachear. Bueno, va bien saberlo, para eso probamos.
Vamos a ver como cachea esto ahora contra contenido estático. Para eso utilizaremos el software de cache de WP Super Cache, una evolución del de R.Galli, que lo que hará es generar .html estáticos. Así que configuré ambos entornos con ese software y me dispuse a probar.
2. Contenido estático
# ab -c10 -n10 http://rubenortiz.es/
This is ApacheBench, Version 2.0.40-dev <$Revision: 1.146 $> apache-2.0
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Copyright 2006 The Apache Software Foundation, http://www.apache.org/
Benchmarking rubenortiz.es (be patient).....done
Server Software: Apache/2.2.3
Server Hostname: rubenortiz.es
Server Port: 80
Document Path: /
Document Length: 0 bytes
Concurrency Level: 10
Time taken for tests: 1.700771 seconds
Complete requests: 10
Failed requests: 0
Write errors: 0
Non-2xx responses: 10
Total transferred: 5400 bytes
HTML transferred: 0 bytes
Requests per second: 5.88 [#/sec] (mean)
Time per request: 1700.771 [ms] (mean)
Time per request: 170.077 [ms] (mean, across all concurrent requests)
Transfer rate: 2.94 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 2 1.3 3 3
Processing: 818 1193 277.8 1235 1696
Waiting: 816 1192 278.4 1235 1696
Total: 820 1195 278.2 1238 1699
Percentage of the requests served within a certain time (ms)
50% 1238
66% 1247
75% 1344
80% 1562
90% 1699
95% 1699
98% 1699
99% 1699
100% 1699 (longest request)
ab -c10 -n10 https://www.rubenortiz.es/
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking www.rubenortiz.es (be patient).....done
Server Software: LiteSpeed
Server Hostname: www.rubenortiz.es
Server Port: 80
Document Path: /
Document Length: 83908 bytes
Concurrency Level: 10
Time taken for tests: 0.012 seconds
Complete requests: 10
Failed requests: 0
Write errors: 0
Total transferred: 843060 bytes
HTML transferred: 839080 bytes
Requests per second: 841.82 [#/sec] (mean)
Time per request: 11.879 [ms] (mean)
Time per request: 1.188 [ms] (mean, across all concurrent requests)
Transfer rate: 69307.25 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 1 2 0.6 3 3
Processing: 2 5 2.1 4 9
Waiting: 0 4 2.4 4 8
Total: 3 7 2.5 7 11
WARNING: The median and mean for the initial connection time are not within a normal deviation
These results are probably not that reliable.
Percentage of the requests served within a certain time (ms)
50% 7
66% 8
75% 9
80% 10
90% 11
95% 11
98% 11
99% 11
100% 11 (longest request)
2.1 Resumen test
Apache 2.2.3 => 5.88
LiteSpeed => 841.82
Wow, ahora si el tema cambia. Como se puede observar, LS sirve muchísimo más rápido el contenido estático. No hay color.
Por otra parte, si vamos aumentando la concurrencia de sesiones que piden la home cacheada de wordpress, veremos que mientras por Apache el resultado es mucho mucho peor, en LiteSpeed aumenta el rendimiento, increíble!
Pros de LS: compatible con Apache, estable, muy buen rendimiento en contenido estático.
Es una opción a tener en cuenta en el futuro a corto plazo 🙂
