Website Builder Global Content Delivery Network

Today we have launched our new global content delivery network (CDN).

Our CDN delivers your websites to your end users from the closed node. We serve your website content from 4 separate locations around the world to ensure the fastest download times.

Our international web server locations are

  • Auckland, New Zealand
  • Sydney, Australia
  • California, USA
  • Ireland, Europe

The benefits of our CDN are that your website can be downloaded faster by users around the world. Rather than coming back to a single location in NZ, a user in London only needs download content from Ireland. That may only be a matter of seconds, but for pages with lots of images, that time adds up. 

The only complication of a CDN, is that publishing of new content may take up to 30 seconds during busy times. Please wait 30 seconds after publishing your website via our CMS before worrying about content not publishing to your live website. Please let us know if you do have any publishing issues that take longer than 30 seconds. 

Some other benefits of our CDN service, is that if 1 node goes down, the other nodes will automatically take over. Recently we had a major hardware failure of a crucial component of our system, and this caused about 50% of web traffic to fail for 21 minutes. That failure was a big wake up call, and it pretty much was a real world test of a worst case scenerio. If the same issue happened today, the outage would only have effected only 25% of traffic for a maximum of 10 minutes, and it would have resolved itself without human interaction, by droppnig the failing node. 

All our customer websites are now dynamically served in the following way. 

  1. Four different DNS servers around the globe will provide an IP address for the closest web server in the fastest possible time. 
  2. Multiple web servers, in four different geographical locations will serve that content in the fastest possible time, being the server closest to the user. If one node fails for any reason, the other nodes will automatically take over. 
  3. 2+ different application servers will serve the dynamic requests, such as site search and shopping carts. If one machine fails, the other will take over. If server load increases, more application servers will make themselves available to handle requests automatically. 
  4. Our core database is hosted in a multi A/Z environment. Meaning that we have a 2nd copy of our database on warm standby should the primary instance fail. This is an extremely unlikely event, but should it occur, our backup database will be up and running within seconds or minutes with very little loss of data. Most importantly, the redundant machine will be up and running without human intervention, day or night. 

Please compare the above infrastructure with your typical smaller web hosting company. A typical small open source web hosting company will host hundreds or thousands of websites on a single virtual private server (VPS). If there is a failure of the VPS, or more likely, a failure of your database, it may require human intervention to resolve the issue. Perhaps your web designer is asleep, or on a plane, or a bus. Perhaps she/he needs to restore your database from backups. This will be doable, but it might be 1+ hours before they are working on the restoration, and another hour before they have restored your data, and restarted their web servers. Your data might be missing a whole day of sales data. 2 hours is a long time to have an outage.

When comparing hosting companies please consider how they will handle your worst case scenerio. Is it better to spend less than 50cents a day for lousy web hosting, or $1 a day for robust web hosting... Both amounts are too cheap for you to consider anything but the very best hosting infrastructure. 

Posted: Wednesday 22 April 2015


No messages found!