New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
OpenLiteSpeed benchmark discussion
Seems other thread's dominant discussion is over Litespeed Enterprise TOS http://www.lowendtalk.com/discussion/10383/litespeed-web-server-goes-open-source-with-openlitespeed. So starting this thread for folks to post their own OpenLiteSpeed benchmarks and results from their own testing
Comments
Only started testing but looks good so far, stock out of box Centmin Mod Nginx v1.4.1 vs stock of out box OpenLiteSpeed 1.0 for static files. Of course concurrency levels are low so not enough to differentiate between the 2. Only changes made is set Nginx worker processes to 4 and same for OpenLiteSpeed.
Seems OpenLiteSpeed wins on Longest Transaction time being faster.
Note default Centmin Mod Nginx also has open_file_cache enabled as well.
Siege Benchmark v3.0 Tested against
static nginx-logo.png default logo included with Nginx installs
Nginx on port 80
Nginx hello.txt
OpenLiteSpeed hello.txt
Nginx nginx-logo.png
OpenLiteSpeed nginx-logo.png
-----------------------------------------
Server Configuration:
-----------------------------------------
VirtualBox - CentOS 6.4 64bit
Xeon W3540 @3.2Ghz allocated 4 cpu threads
1280MB memory
20GB disk on 1TB Western Digital Black Caviar
-----------------------------------------
Software:
-----------------------------------------
Centmin Mod v1.2.3-eva2000.01 Beta http://centminmod.com/centminmod_v123.html
Nginx 1.4.1 with ngx_pagespeed support turned off
PHP 5.2.4 PHP-FPM + APC Cache 3.1.13 + Memcache 3.07 + Memcached 2.1.0 + Libmemcached 1.0.16
MariaDB 5.2.14 MySQL
OpenLiteSpeed 1.0
PHP 5.3.24 LSAPI + APC Cache 3.1.13 + Memcache 3.07 + Memcached 2.1.0 + Libmemcached 1.0.16
-----------------------------------------
Siege Benchmark 3.0 settings
-----------------------------------------
- Cache validation disabled (default)
- Concurrency levels tested: 50, 100, 150, 200, 250
- Reps: 10
- Delay: 5 seconds
- Siege benchmark mode: bench
Nginx nginx-logo.png
OpenLiteSpeed nginx-logo.png
Actually the other thread is now about minorities, murder, and religion. Thanks for sharing your benchmarks like a normal person does when talking about a web server. I'm quite interested in learning what the benefits are of LiteSpeed without so many of the key features.
Did you expect anything better?
Yeah I'm just interested in performance of OpenLiteSpeed particularly in comparison to Nginx. I chose lowendtalk for the discussion as on average you folks would have a higher technical expertise for this compared to other forums as well. That and alot of pro-Nginx folks
Let's see main features missing from OpenLiteSpeed vs Litespeed Enterprise
Hosting control panels - how many Nginx users even use a control panel ?
.htaccess file compatibility - it's only the file not supported, Apache rewrite syntax is fully supported but seems they moved that up into vhost admin console configuration in rewrite and context tabs. Haven't played with that aspect of OpenLiteSpeed yet but at least it uses Apache rewrite syntax !
mod_security, how many folks use mod_security ? for Nginx ?
page caching - I assume this means Litespeed Enterprises' inbuilt Litespeed Cache. That would probably be the feature I'd miss the most. I wrote about Litespeed Cache vs Varnish Cache 2+ yrs ago on my blog http://vbtechsupport.com/33/. Litespeed Cache performed as good if not better than Varnish cache for dynamic PHP caching.
But OpenLiteSpeed still retains Litespeed Enterprises inbuilt small static file and MMAP cache support from the admin console screenshot (see below). So static file cache/performance is right up there with Varnish Cache for static files (in theory).
@eva2000 I really noobs about OpenLiteSpeed. Could you write some tutorial on installing wordpress.
I see only one problem with LiteSpeed webserver.
It's enterprise edition, what cost money for get all features.
Nginx give a lot features for free from box. Same as like apache.
I hope, Nginx and LiteSpeed will be good concurents, and they remove monopoly from Apache by their features / quality / speed / resource usage.
it's free? Oh more spam site / autoblog.
Thank you for taking the time to post your results eva2000. Appreciate it!
@eva2000 One of the main reasons I shifted from Apache to Nginx on my LEBs was due to high memory consumption. Interested in knowing how does OpenLiteSpeed compare with Nginx in terms of memory consumption (right from zero concurrency)?
@Noerman, if i have time I may eventually - seems LitespeedTech are planning guides for common web scripts too so might keep an eye out for that on their web site, blog and google group
@Noxter you're welcome
@vanarp, early tests show memory usage very close between the 2 of course depends on specific Nginx or OpenLiteSpeed settings used. But for Centmin Mod based Nginx vs OpenLiteSpeed very close. Siege benchmarks show Nginx peak 150-210MB memory usage vs OpenLiteSpeed 150-180MB memory peak and idle well
root 19603 0.0 0.0 58052 1008 ? S 00:35 0:00 _ openlitespeed (lscgid)
nobody 19604 0.0 0.1 59116 1808 ? S 00:35 0:00 _ openlitespeed (lshttpd - #01)
nobody 19605 0.1 0.1 59116 1808 ? S 00:35 0:00 _ openlitespeed (lshttpd - #02)
nobody 19606 0.1 0.1 59116 1808 ? S 00:35 0:00 _ openlitespeed (lshttpd - #03)
nobody 19607 0.0 0.1 59116 1808 ? S 00:35 0:00 _ openlitespeed (lshttpd - #04)
root 20572 0.0 0.1 41728 1396 ? Ss 00:35 0:00 nginx: master process /usr/local/sbin/nginx -c /usr/local/nginx/conf/nginx.conf
nginx 20573 0.1 0.2 42716 2604 ? S< 00:35 0:00 _ nginx: worker process
nginx 20574 0.1 0.1 42164 2104 ? S< 00:35 0:00 _ nginx: worker process
nginx 20575 0.2 0.1 42164 2104 ? S< 00:35 0:00 _ nginx: worker process
nginx 20576 0.2 0.1 42164 2088 ? S< 00:35 0:00 _ nginx: worker process
I've been meaning to do a simple benchmark comparison - I'm trying to think of the best environment to do it in with http://loader.io/ - maybe AWS? I have DO credit but I don't trust their VMs
Only real problem is you need to test at very high concurrency based loads to see Litespeed (assume OpenLiteSpeed) benefits over Nginx. So target test server needs to be able to handle it. As such I'm moving my test from virtualbox to a more beefier server(s) eventually
@eva2000 And also testing concurrency on a typical WordPress site will give much better idea since most sites would be running WP anyway IMHO.
Thanks @eva2000.
I may have found an apache replacement
heads up official quick install guides for wordpress, joomla and magento are up on their official site http://open.litespeedtech.com/
You are awesome! I am excited to try it for WP over the weekend.