Not Seeing any Speed Improvement

[sagesmart]

Hi,

I recently installed LiteSpeed on my server and everything seems to be working okay. However, I compared the page load times of one of my sites both before and after installing LiteSpeed and I see no difference at all. I used the YSlow Firefox plugin to quantify page load times.

Prior to installing LiteSpeed I was running PHP 5.3.6 with suPHP. I used the LiteSpeed WHM plugin to "Build Matching PHP Binary". Also, during the LiteSpeed installation I made sure to indicate that PHP suEXEC is set to YES.

Also, PHP suEXEC Max Conn is set to 10 and Priority is set to -19.

Either something isn't configured correctly, or my expectations of what the performance improvements would be were not based on reality.

The website I'm testing is a Wordpress website with no caching at all. I intentionally haven't cached anything yet because I wanted to see how much improvement was gained without caching before going any further.

Both before and after, the YSlow plugin is showing an initial server response time of 300 to 350ms. And then it takes a few hundred more milliseconds to finish loading all the rest of the stuff on the page (e.g. style sheets, images, etc.). I was expecting something more along the lines of 50ms.

Any ideas on why I might not be seeing the improvement I thought I'd see? Considering I was previously using Apache with suPHP I was expecting a drastic and noticeable improvement, but so far that hasn't been my experience.

 

[mistwang]

You will see bigger difference if your server is under heavy load. Network latency is playing a role here as well.

PHP script itself will run at the same speed inside PHP engine, LiteSpeed's advantage is resource management, PHP startup, passing request/response, etc. The faster your scripts run, the bigger difference you will see.

Under heavy load, Apache's performance will degrade greatly, while litespeed will hold much better.

If you want, you can benchmark a "Hello World" php script from localhost with "ab".

 

[webizen]

maybe the connection between your test machine and the server in question is slow.

... And then it takes a few hundred more milliseconds to finish loading all the rest of the stuff on the page (e.g. style sheets, images, etc.). I was expecting something more along the lines of 50ms.

...

 

[sagesmart]

Thanks for the replies. I sounds like my expectations were not based on reality. In addition to being better (than Apache) at managing a large number of simultaneous connections I guess I was also expecting to see similar improvements even with a single connection.

If I understand you correctly, it appears that PHP is my bottleneck, not necessarily LiteSpeed. So, perhaps eAccelerator along with LiteSpeed will give me the performance I'm looking for with my Wordpress site.

 

[webizen]

...
If I understand you correctly, it appears that PHP is my bottleneck, not necessarily LiteSpeed. So, perhaps eAccelerator along with LiteSpeed will give me the performance I'm looking for with my Wordpress site.

That's right. LiteSpeed Enterprise edition has Cache built-in (aka page cache the output instead of opcode cache) which greatly reduce PHP invocation and/or PHP binary code execution as well as DB access.

Here is a guide to LSWS cache:

http://www.litespeedtech.com/support/wiki/doku.php?id=litespeed_wiki:litespeed:cache

 

[cmanns]

Thanks for the replies. I sounds like my expectations were not based on reality. In addition to being better (than Apache) at managing a large number of simultaneous connections I guess I was also expecting to see similar improvements even with a single connection.

If I understand you correctly, it appears that PHP is my bottleneck, not necessarily LiteSpeed. So, perhaps eAccelerator along with LiteSpeed will give me the performance I'm looking for with my Wordpress site.

Try XCache & W3 Total Cache, maybe add Memcached but ususally W3, XCache and proper PHP tuning do the task.

Drop me a PM if you need some serious tuning, we host extremely high traffic (Dozens of php requests PER Second) on the same system, so if your pushing less than a 100 or so tops and bottlenecking, think about a system with over 16gigs of 2-4mil post forums and busy WordPress's all getting along happily 24/7/365 monitored or I loose $$$$