Look, we all know it’s been a tough year. A divisive presidential campaign; celebrity deaths including David Bowie, Prince and Leonard Cohen; the Syrian refugee crisis; the Zika virus.
So here’s a bit more bad news: this year is going to be one second longer than most.
You knew that 2016 is a leap-second year, right?
This is the 27th leap-second year since 1972, when the Bureau International de l’Heure undertook to remedy the recurring but irregular gap between Coordinated Universal Time and mean solar time. Without such corrections, the time based on the (somewhat erratic) elliptical orbit of the Earth would drift away from (extremely consistent) atomic time. That could cause big problems with time-sensitive technology, such as GPS.
In general, the leap second will be inserted immediately after 23:59:59 on Dec. 31. That means the last minute of 2016 will contain an extra second, which is fine for most purposes.
But in computing, transactions and processes can be closely linked to time, and programming errors involving leap-days and leap-seconds can wreak havoc, causing crashes, unexpected interface behavior, and errors. They can also affect some forms of encryption that embody time-based calculations. Microsoft Azure catalogued and candidly addressed these issues earlier this year, admitting it had had a leap-year outage in 2012 and reminding users that 2016 has 366 days, which could cause unexpected issues Dec. 30 through Jan. 1.
So public-cloud giant Amazon Web Services is planning to deal with the leap second in a different and somewhat intricate manner.
Aws will spread the extra second over the 12 hours before and the 12 hours after the leap second. During that 24-hour period, Amazon will add 1/86,400 of a second to each second occurring between just before noon (11:59:59) on December 31, 2016, and noon (12:00:00 ) on January 1, 2017. “AWS Adjusted Time and Coordinated Universal time will be in sync at the end of this time period,” according to a blog post. More of the bloody details are presented here.
If last year’s advice is any guide, that fix will be fine for the AWS Management Console and back-end systems and for instances of Windows running on Amazon machine images, which will follow AWS Adjusted Time. It’s irrelevant to EC2 instances, which have user-controlled clocks.
The resolution sounds a little more problematic for some AWS services and most Amazon Relational Database instances, which will show “23:59:59” twice. If a transaction occurs during that two-second period, how will staffers or analysts be able to tell at which of the two 23:59:59’s it occurred?
Some AWS versions of Oracle will follow AWS Adjusted Time. Others won’t.
Microsoft doesn’t specifically address how Azure will address the coming ;eap second, though it’s deep and thoughtful on coping with the leap year. Here’s how Google handled it.
But this isn’t the first time either cloud service has gone through a leap-second event. So even though there’s a chance the extra second could foul up cloud computing, the odds seem small.
Regardless of which cloud you might use, there’s one thing most of us can be glad about: One second is too short for a certain someone to type a tweet.
