Well, its not about any special host or operating system, just about the pure performance benchmarks measured for distinct setups of software on some kind of hardware.
Setups:
- Apache 2.0.52, PHP 5.0.3, Mysql 4.1.7
- Lighthttpd 1.3.5, PHP-CGI 5.0.3, spawn-fcgi 1.2.0
- both configured for PHP memory_limit req.
I was testing with Siege the server on localhost with permissions.
Copy to clipboard
siege -c 10 -t 2m http://tiki19/tiki-index.php
|
Results with Lighthttpd and PHP5-FCGI <br />
700Mhz, 384MB RAM, udma2 ides ext3 fs<br />
Transactions: 103 hits<br />
Availability: 100.00 %<br />
Elapsed time: 120.26 secs<br />
Data transferred: 2164266 bytes<br />
Response time: 11.02 secs<br />
Transaction rate: 0.86 trans/sec<br />
Throughput: 17996.56 bytes/sec<br />
Concurrency: 9.43<br />
Successful transactions: 103<br />
Failed transactions: 0<br />
Longest transaction: 14.61<br />
Shortest transaction: 1.91<br />
|
Results with Apache2 PHP5 <br />
700Mhz, 384MB RAM, udma2 ides ext3 fs<br />
Transactions: 69 hits<br />
Availability: 100.00 %<br />
Elapsed time: 119.96 secs<br />
Data transferred: 1449585 bytes<br />
Response time: 15.70 secs<br />
Transaction rate: 0.58 trans/sec<br />
Throughput: 12083.90 bytes/sec<br />
Concurrency: 9.03<br />
Successful transactions: 69<br />
Failed transactions: 0<br />
Longest transaction: 31.38<br />
Shortest transaction: 4.97<br />
|
No php caching was used, cpu load was much lower with phpfcgi, running a whole X desktop with Firefox and Evolution and the like in the background...
For some more benchmarks on Lighthttpd go here.
;){kind=link}
;){kind=link}