August 27, 2013 Case Study, General


We’re developers, not sysadmins. Our previous hosting company kept on telling us to add more boxes. We did, but later discovered that they’d never been connected to the load balancer!

Working with Anchor has been a relief — extremely professional and very experienced. They reconfigured with fewer servers, better site performance, and we have a clear growth plan.”

Ben Still – Managing Director, Redant

The Company

Huggies is just one of the many brands in the Kimberly-Clark portfolio. The Huggies website houses what is likely the world’s largest and dynamic online community for parents. Who would have thought that the parents of the world would come together thanks to a nappy manufacturer? The forum on the Huggies website alone has seen over 2.6 million messages posted by almost a million users. That’s close to one in every 22 Australians, and closer to one in ten parents — a big number for any organisation!

messages posted in the Huggies forum

The Opportunity

Anchor was introduced by Kimberly-Clark’s web development partner, Red Ant. They challenged us to design, build and implement an infrastructure to support the massive volume of traffic that the Huggies website receives on a daily basis.

While the sheer volume of traffic was a major consideration, other aspects of the deployment were also critical considerations in the design of the server architecture. Much of the site’s content would be delivered using Ruby on Rails behind an Nginx webserver, while the forums were PHP-based and required a MySQL backend. If that wasn’t challenging enough, the scope included legacy applications that could only run on Windows, necessitating a hybrid Windows/Linux environment where both operating systems could co-exist harmoniously.

The most important aspect of this deployment was high reliability — a great deal of damage could be done to the Huggies brand if consumers weren’t 100% satisfied with their experience of the website. It was essential that there were no single points of failure anywhere in the design.

The Anchor Solution

This brief gave us much to consider, but that’s what we love! The Anchor team really enjoy complex problems, getting the opportunity to put on our thinking caps and coming up with innovative solutions.

Since reliability was so critical, our design ensured redundancy was built in to the solution at all levels, from the upstream provider’s network, through the switching layer, down to the individual servers themselves.

Server-grade rackmount hardware was selected, but given the high uptime requirement we provided two servers for each role, further enhancing reliability. To comfortably manage high volumes of parenting data, traffic would be balanced between multiple frontend servers, using a high-availability load-balancer.

For the backend, the database was configured in an active/passive mode, with automatic failover in the event of hardware or network failure. For data consistency between nodes, we used Xen virtualisation and DRBD (Distributed Replicated Block Device), a design that works equally well in supporting Linux and Windows servers.

With this solution design we knew that we had something that was absolutely rock solid, could handle the existing traffic loads, and was capable of being scaled horizontally in future if necessary.

Now the parents of Australia can be assured of an answer to their early childhood questions.