My inner benchmark demon took over and got a bit side tracked from testing out Litespeed's inbuilt cache feature. LOL
I ran these static test.txt file apachebench benchmarks to rule out an a possible server config/environment issue which showed up in cache testing in cache tests part 2 - particularly was alot worse with nginx for test.php tests like apache with test.php file details.
Even with static test.txt file, repeated back to back apachebench runs resulted in lower requests per second for apache. While NginX didn't exhibit as bad a result compared to it's test.php tests (2nd run already down to 330 rps). Litespeed seemed immune to dramatic drops at least.
Note:
I ran these static test.txt file apachebench benchmarks to rule out an a possible server config/environment issue which showed up in cache testing in cache tests part 2 - particularly was alot worse with nginx for test.php tests like apache with test.php file details.
Even with static test.txt file, repeated back to back apachebench runs resulted in lower requests per second for apache. While NginX didn't exhibit as bad a result compared to it's test.php tests (2nd run already down to 330 rps). Litespeed seemed immune to dramatic drops at least.
Note:
- CPU and Memory utilisation were much higher for Apache than NginX than Litespeed. For example, by 5th ran on 1000 concurrency level, Apache cpu load was hitting 2.8, NginX was around 1.2, and Litespeed around 0.6.
- test.txt file used contents was ascii art.
- NginX set to 2 workers to compare to 2 cpu Litespeed license
- Raw apachebench numbers available here
![](http://vbtechsupport.com/litespeed/benchmarks/jan23/nginx0854/ipconntrack_tuned/apachebench_static_textfile_apache_nginx_litespeed_01.png)
Last edited: