Amazon said that the downtime encountered on Sunday by its European customers was caused by hardware failure than by a barrage of distributed denial of service, something that might prove to have a negative impact on how cloud computing is perceived.
The Amazon UK, DE, FR, IT and AT web stores suffered around 30 minutes downtime yesterday night and were all located in the company's data centre in Ireland which also houses servers for Amazon's EC2 service in the EU region.
Amazon's EC2 service level agreement mentions a 99.95 per cent availability which equates to around 270 minutes (four and a half hours) downtime per annum for each region, which means that downtime in each region are not aggregated.
While Amazon's cloud computing structure appears to be pretty much resilient when it comes to classical security threats like DDoS, it appears to be limited when it comes to handling hardware failures.
And as more services move to giant server farms, the risk that one small glitch takes down a data center rises exponentially because of the number of variables involved.
We suspect that Amazon's outage might share some commonalities with the downtime that affected Gmail nearly two years ago where a scheduled routine maintenance caused a data center to overload and cause a chain reaction like falling dominos.
Some might remember that Amazon.com had some serious downtime towards the end of June 2010, something which didn't appear to affect Amazon Web Services, just like the one of yesterday night.