Swine flu lessons: Avoiding website failures

On the 23rd of July 2009, at 3pm, the Government's National Pandemic Flu Service website launched, providing a self-assessment service which the public could use to see if they had got the symptoms of swine flu. At 3.04pm, the website crashed after several thousand people every second attempted to access the website.

Whilst there was certainly unprecedented demand to access the site, many critics expressed surprise that the Government failed to build in the capacity to cope with the extra visitors to such a high-profile site. In this day and age, one might think that it is unacceptable for public information websites to go down when people need them the most, especially given swine flu’s undeniably high profile.

Despite the technology and planning tools available today though, many important websites still grind to a halt or simply go down when they have an unanticipated number of visitors. Many e-commerce websites continue to crash during busy sales or at Christmas, whilst the National Rail website crashed during the heavy snow earlier in the year, when people logged on to see whether they would be able to get to work by train.

The common solution to this challenge is to add extra server capacity, or to split the traffic across multiple servers, rather like a football ground strategically channelling fans into and out of the stadium through different entrances and exits.

Whilst a significant number of website failures are caused by unexpected visitor numbers, they can also fail if they aim too high and try to do too much without having adequate technology in place.

Websites hosting long videos, music for download or large image files can be overwhelmed if people try to download too much at once. However, in this instance the NHS flu pandemic website has an extremely small footprint - there are around ten pages in total, none of which host large images or multimedia files. It is estimated that the site was prepared for around 1,200 users per second, but received over twice that in its opening minutes – over 600,000 users in 4 minutes (approximately 1% of the population of the UK).

The upshot of all this was that people were unable to take the tests on the site until the Government had dealt with the problem, potentially causing panic or placing more pressure on NHS phone operators. However, to put the problem in perspective, the government had anticipated 0.5% of the UK population would log onto the website in the first four minutes – around seven times the volume which the BBC News website receives.

That said, with the advent of virtualised Cloud infrastructure available on an on demand, ‘pay as you use’ basis, one wonders why this site couldn’t cope with any demand thrown at it.

Yes this would have cost extra money, but surely that extra investment would have been worth it to ensure that such a vital service didn’t go down. By using the new generation of hosted technologies which effectively deliver ‘burstable’ infrastructure, the Government could have had a website that would have easily coped with these unprecedented levels of traffic.

So how can we cope with the expected? From a hosting perspective the answer is clear: build ‘on demand’ infrastructure that can cope with any load. For those who are building public information sites, or any site that is likely to see high traffic demand for a short period of time, the on demand model is something they should pay close attention to. It will give them access to all the capacity they need, but ensure that they only pay for what they use.

Ultimately, all those responsible for running important websites need to stand up and take note. If you don’t want to risk losing your customers or letting the public at large down, you must ensure you’ve got the capacity available to cope with unpredicted levels of interest in them.

Neil Barton, is a director of web hosting company Hostway UK

Copyright © 2009 IDG Communications, Inc.

Shop Tech Products at Amazon