Help Websites Planning Audience Peaks

Planning Audience Peaks

Discover our Public Cloud offer

Shared Hosting for Everyone, imagined by developers, for developers.

Discovering the Public Cloud

You are planning a TV/Radio presence that may cause more visits and you are not sure that your site can take the load? Here are some actions you can take ahead of time:

  • Isolate the site by account - to avoid interfering with other sites as resources are shared at the account level,
  • Enable the HTTP cache on your administration interface and/or at the application level using Memcached, Redis 1 or any other software cache,
  • Change to the latest possible language versions - to improve performance,
  • Set a redirect to a static page - using far less resources, these can be sent faster.

To observe the behavior of your application you can also run a load test (e.g. using ab).

Example avec ab

We want to see if our commercial website could handle 300 000 visits per day.

The following test makes 100 requests on the home page at a rate of 10 at a time (in parallel).

$ ab -c 10 -n 100 https://www.alwaysdata.com/en/
This is ApacheBench, Version 2.3 <$Revision: 1843412 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/

Benchmarking www.alwaysdata.com (be patient).....done


Server Software:        nginx
Server Hostname:        www.alwaysdata.com
Server Port:            443
SSL/TLS Protocol:       TLSv1.2,ECDHE-RSA-AES128-GCM-SHA256,2048,128
Server Temp Key:        X25519 253 bits
TLS Server Name:        www.alwaysdata.com

Document Path:          /en/
Document Length:        3774 bytes

Concurrency Level:      10
Time taken for tests:   7.347 seconds
Complete requests:      100
Failed requests:        50
   (Connect: 0, Receive: 0, Length: 50, Exceptions: 0)
Non-2xx responses:      50
Total transferred:      4007050 bytes
HTML transferred:       3961700 bytes
Requests per second:    13.61 [#/sec] (mean)
Time per request:       734.658 [ms] (mean)
Time per request:       73.466 [ms] (mean, across all concurrent requests)
Transfer rate:          532.65 [Kbytes/sec] received

Connection Times (ms)
              min  mean[+/-sd] median   max
Connect:       82  121 114.3    100    1114
Processing:    33  521 813.1    212    4216
Waiting:       33  458 800.5     96    4094
Total:        120  642 816.0    320    4318

Percentage of the requests served within a certain time (ms)
  50%    320
  66%    638
  75%    813
  80%    878
  90%   1480
  95%   2318
  98%   4313
  99%   4318
 100%   4318 (longest request)

The important information is: Requests per second: 13.61 [#/sec] (mean).

300,000 visits in 10 hours (small day) corresponds to 8.3 requests per second. The current configuration is therefore compatible.

  1. These two applications are available on Private Cloud environments. ↩︎