This is a brilliant example of 2 key Azure workloads: predictable-bursting and unpredictable-bursting. In the horse caring calendar there are big races such as The Grand National and Royal Ascot (that’s the one where everybody gets dressed up and outrageous ladies hats are most in evidence).
Do you recognise the hats? Quality Street Chocolates…
These are totally predictable events and At The Races have enough historical data to know when and by how much to start the scale-up of their site. But there are also unpredictable events such as the death of a famous race-horse which will result in many visits to the site. At The Races are also able to cope with this load easily simply by scaling the site to match the demand. It means as far as their web-visitors are concerned the site is always responsive.
There is a case study that discusses why they moved to Azure.
I made a video of their first release on to Windows Azure some time ago:
It’s well worth reading the case study and watching the video if you run a site that suffers a similar business pattern of predictable and unpredictable peaks.