Measuring Amazon AWS Uptime

Yesterday, for the second time in two years Techblog was offline most of the day due to an Amazon AWS service outage. That’s effectively, for business purposes, 364/365 = 99.726% uptime.

99.726 percent uptime isn’t a statistic we hear about (returns zero results on Google) because cloud hosting providers shine up the numbers the same way mutual fund managers spin cumulative investment returns. For example, yesterday Amazon’s Virginia datacenter wasn’t actually down for a full twenty-four hour period, and not every datacenter on their global network was affected.

Even with yesterday’s elastic block storage (EBS) service outage, Amazon can likely still claim high availability for most applications — though for R/GA’s Techblog team, it feels like a bit less. And now that the inconvenience is over, there are things to be thankful for. The pros at Amazon fixed the problem, made our application reappear and Techblog was back in action with no data loss — just like last year.