Microsoft has been running its Azure cloud services since 2009, touting the technology as a robust and secure way to built and deploy cloud-based applications, Web sites, and other technology that can be available to users anywhere they can get Internet access — and without having to worry about complicated stuff like VPNs and secure tunnels back to corporate HQ. However, on Wednesday the service experienced a 12-hour outage thanks to one of the most embarrassing bugs a programmer can make: Azure didn’t handle February 29 leap day correctly.
Without noting the specific problem, Azure lead engineer Bill Laing noted in a blog post that the outage was determined to have been caused by a “time calculation that was incorrect for the leap year.” Data Center Knowledge’s Rich Miller reported that Microsoft attributed the problem to a “cert issue” — which would likely mean that certificated used to establish secure communications and access between related systems failed to validate correctly on February 29. Liang apologized for the outage and the inconvenience to customers, and promised a full analysis would be published within ten days.
The incident serves to highlight the essential problem with all cloud services: every once in a while, they will treat their customers to a bright, sunny, cloudless day. Back in April, Amazon’s cloud services suffered a sustained crash that resulted in the loss of customer data; in August both Amazon and Microsoft’s cloud services were taken offline in Europe by a power failure in Dublin, Ireland.
[Azure advertisement: Microsoft]