Travis Mijat's questions

I used ab (ab -c50 -t5s) to benchmark Apache V Nginx for a static file on the same server.

Apache becomes CPU bound at 200 requests per second and Nginx is at about 25% CPU while handline 5211 requests per second.

No swapping took place with Apache, lack of memory wasn't an issue.

I know there are a myriad of different configurations for both Nginx and Apache, but how far behind Apache is Nginx is just ridiculous.

So ridiculous in fact that I think I've missed something with Apache.

Any ideas where to start?

Here are the ab results:

Server Software:        nginx  
Server Port:            80  

Document Path:          /images/15.jpg  
Document Length:        15088 bytes  

Concurrency Level:      50  
Time taken for tests:   5.494 seconds  
Complete requests:      26059  
Failed requests:        0  
Write errors:           0  
Total transferred:      400370476 bytes  
HTML transferred:       393178192 bytes  
Requests per second:    5211.29 [#/sec] (mean)  
Time per request:       9.595 [ms] (mean)  
Time per request:       0.192 [ms] (mean, across all concurrent requests)  
Transfer rate:          78189.48 [Kbytes/sec] received  

Connection Times (ms)  
              min  mean[+/-sd] median   max  
Connect:        0    4   0.6      4      10  
Processing:     3    5   0.8      5      13  
Waiting:        0    2   1.3      2      10  
Total:          6    9   0.9      9      19  

Percentage of the requests served within a certain time (ms)  
  50%      9  
  66%      9  
  75%      9  
  80%      9  
  90%     10  
  95%     10  
  98%     11  
  99%     12  
 100%     19 (longest request)  


Server Software:        Apache  
Server Port:            81  

Document Path:          /images/15.jpg  
Document Length:        15088 bytes  

Concurrency Level:      50  
Time taken for tests:   5.31551 seconds  
Complete requests:      1008  
Failed requests:        0  
Write errors:           0  
Total transferred:      15681815 bytes  
HTML transferred:       15433585 bytes  
Requests per second:    200.34 [#/sec] (mean)  
Time per request:       249.581 [ms] (mean)  
Time per request:       4.992 [ms] (mean, across all concurrent requests)  
Transfer rate:          3043.59 [Kbytes/sec] received  

Connection Times (ms)  
              min  mean[+/-sd] median   max  
Connect:        2   84  47.5     78     222  
Processing:    24  158  46.5    162     260  
Waiting:        1   96  45.1     91     221  
Total:        187  242  48.5    232     428  

Percentage of the requests served within a certain time (ms)  
  50%    232  
  66%    238  
  75%    246  
  80%    249  
  90%    304  
  95%    380  
  98%    409  
  99%    423  
 100%    428 (longest request)  

Here is an example shell script I'm looking at implementing using inotify:

inotifywait -mrq -e close_write -e create -e delete -e move /var/www | while read file;
do   
    rsync -av /var/www1/ /var/www2/  
done

Let's say rsync takes 30 seconds to complete.

If there are 10 inotify events within 5 seconds, will there be 10 processes of rsync running the same job at the same time?

OR

Will there only be one rsync job that is run 10 times, one after the other?

I'm planning to share a GFS2 filesystem via iSCSI to my web servers.

Are the files read by my web servers cached locally using the linux disk cache stored in memory or are they only cached on the GFS2 cluster?