Amazon’s Cloudfront DNS goes cold turkey on eve of Thanksgiving
It was only a couple of weeks ago that Microsoft’s Azure was hit by an extended global outage that took its cloud services offline for several hours. Now, just before Thanksgiving, Amazon Web Services Inc. has suffered a similar fate.
The problem hit Amazon’s Cloudfront DNS service late yesterday afternoon, causing a major slow down in page load times for numerous websites that use the service, including sites like Instagram and Medium.
A notice was posted to AWS’ status page admitting the problem:
“We are currently investigating increased error rates for DNS queries for CloudFront distributions,” it said at around 5:00pm PST. Luckily the outage didn’t last too long, and by 5.57pm the page was reporting “Error rates for DNS queries of CloudFront distributions are currently recovering.”
Cloudfront was back up and running normally as of 6.02pm, according to Amazon’s status page.
But it was a major problem that went on for over one hour and was seen as far away as Australia. Posters to the AusNOG mailing list actually reported problems lasting as long as two hours, before Amazon re-routed traffic through different undersea cables to fix the issue.
The outage is somewhat reminiscent of Amazon’s Elastic Load Balancer error on Christmas Eve 2012, which managed to take Netflix offline just as thousands of families sat down to enjoy some holiday TV entertainment.
With Americans on the verge of whipping out their credit cards for Black Friday, this outage isn’t exactly the best timed one either. Luckily it’s been fixed, but if the problem repeats it could lead to some major frustrations for those turkey-filled, bargain-hunting hordes of online shoppers.
