## Benchmark |
|
In this benchmark, compared two technologies: CppCMS and PHP. |
In this benchmark two technologies were compared: CppCMS and PHP. |
|
This blog, written using CppCMS technology was compared to WordPress blog powered by PHP5. |
|
## Setup |
|
I had compared two blog systems: this one and WordPress 2.5 with a patched WP-Cache-2 addon. I used following configuration: |
|
1. Web Server lighttpd 1.4.13 |
2. Interface FastCGI |
3. PHP 5.2 |
4. Bytecode cacher: XCache 1.2.1 |
5. Database MySQL 5.0 |
6. Caching for WP: WP-Cache-2 with an additional [performance patch](http://art-blog.no-ip.info/cppcms/blog/post/20) that caches already _deflated_ versions of page in order to reduce requirement of re-compressing on cache hit. |
7. Hardware: AMD Athlon XP 64bit, 1G RAM |
8. OS: Linux, Debian Etch 64bit. |
|
I prepared two blogs that were filled up with 1000 articles each. Each article had 10 comments, all the articles were organized in 10 categories in each blog. |
|
## Tests |
|
First I run load tests with disabled caching system. |
|
Then the cache was enabled and cleaned before each test run. |
Each time, the cache was "warmed up" with 100 requests of different pages. |
Then CMS was loaded by http\_load with 1000 requests from 5 concurrent connection. The client was patched in order to send a header: `Accept-Encoding: gzip, deflate` in order to fetch compressed pages: the real live situation. |
|
For each run, the cache included a certain percent of new pages (in order to achieve a correct hit/miss ratio) and the rest where taken from the 100 "warm pages". |
|
## Results |
|
<table> |
<tr><th>CMS</th><th>No cache fetches per second ratio</th><th>Warm up time</th></tr> |
<tr><td>WordPress: </td><td>7.6</td><td>13.0 s</td></tr> |
<tr><td>CppCMS: </td><td>310</td><td>0.28 s</td></tr> |
</table> |
|
<table> |
<tr><th>Miss ratio (%) </th><th> WordPress </th><th > CppCMS </td></tr> |
<tr><td>0</td><td>711</td><td>2300</td></tr> |
<tr><td>1</td><td>370</td><td>2160</td></tr> |
<tr><td>3</td><td>176</td><td>1940</td></tr> |
<tr><td>5</td><td>118</td><td>1790</td></tr> |
<tr><td>10</td><td>64</td><td>1450</td></tr> |
<tr><td>15</td><td>41</td><td>1210</td></tr> |
<tr><td>20</td><td>33</td><td>1075</td></tr> |
<tr><td>30</td><td>20</td><td>795</td></tr> |
<tr><td>40</td><td>13</td><td>570</td></tr> |
<tr><td>50</td><td>13</td><td>570</td></tr> |
</table> |
|
->  <- |
|
## Conclusions |
|
The caching system, significantly improves both CMS written in PHP and C++. However, when hit/miss ratio grows to few percents, the performance of the CMS written in PHP drops very fast and at 3% miss ratio the difference between CppCMS and WP becomes the difference of more then 10 times. |
|
The CppCMS system holds lower hit/miss ratio much better then WordPress with WP-Cache-2. |
