What went wrong with Click Frenzy?
It's fair to say that the hype around Click Frenzy turned sour last week. The website went down when millions of Australian shoppers tried to hop on, resulting in a lot of unhappy people and negative publicity.
Many of the online stores, despite not receiving the click-through traffic they expected from Click Frenzy, went down themselves when people hit their sites directly. Some retailers though, like Terry White Chemists (disclosure: they are hosted with Rackspace) planned well, and were able to cope with the huge spike in traffic.
There are now even reports that the passwords of registrants to the one-day sale were exposed on the site.
While Click Frenzy did a great job at generating interest in the lead up to the sale, its hosting decisions left a lot to be desired. With the affordability, capacity and flexibility of cloud hosting, there truly are no excuses for your site to be going down when you get a foreseeable surge in traffic.
Almost all of the problems that happened with Click Frenzy this week came down to IT mistakes. But every single problem was avoidable. Easily avoidable. This was not a failure of technology, but a failure to understand, plan, and use technology properly.
What can an online retailer do to ensure its cloud costs are manageable, while also being able to cope with huge spikes in traffic?
It comes down to three fairly simple factors; three reasons an online store would crash under a heavy load.
Firstly, the site has to have adequate capacity to handle the influx of new visitors. Obviously that is difficult to provision with in-house infrastructure, which is why most e-commerce sites now go with external providers and their cloud solutions. Because you're renting the capacity from a trusted partner, you can tap into it whenever you need to increase that capacity. An online store's main advantage over its bricks rivals is that it is open 24/7 and convenient. If you start closing down because you've chosen the wrong provider, you're eroding that competitive advantage. Whichever cloud provider an online retailer goes with, they need to have enough scale to grow with them, both short and long term.
Secondly, you need to be closely monitoring what is happening in terms of your infrastructure, the site, its applications, and all the different layers. If you're not closely monitoring it, you won't be prepared for high traffic events and you'll inevitably be responding to it after the fact, when it fails, rather than scaling as you see the traffic start to spike. It's almost like preparing for a tornado to hit. The sites that went down last week were definitely not monitoring activity adequately in the lead up to the start of Click Frenzy.
Thirdly, you need to automate as much of your cloud management as you can. This basically means you enable the site to automatically scale up and down the amount of cloud resource and capacity your application needs. Doing this, as well as ensuring you have adequate capacity, along with close monitoring, will mean any sites attempting to do big one day sales like this will not crash under the pressure.
None of this is rocket science in 2012. As I said, there was no failure of technology here. This was a failure to understand, plan and use technology properly - ironically an accusation that e-tailers and technologists usually throw at our traditional retailers.
Mark Randall is the director at cloud hosting provider Rackspace Australia & NZ.