BackWhy we should fuss about a ‘Leap’ second in a Digital Age
In the past, leap seconds have caused outages at popular websites such as Reddit, Yelp and Foursquare
Premium
Mumbai: By mid-1999, users and companies were panicking that the Year 2000 problem, better known as Y2K, would affect hundreds of computers since programs were designed to shorten four-digit years as two digits in a bid to save memory space and hence additional costs.
This made the year 2000 indistinguishable from 1900 to computers. Fortunately, it was a man-made error that was sorted out and there was no digital disaster as predicted, prompting many to term the Y2K phenomenon as a dud.
However, when you’re dealing with the solar system and the universe, things are very uncertain.
The Earth’s rotation, for instance, is unpredictable in the long term. Hence, atomic clocks, or the world’s timekeepers, have to be adjusted periodically by a leap (one) second.
There have been 25 such adjustments in the last 43 years to ensure that they are in sync with the irregular changes in the Earth’s rotational speed, thus bringing them in line with our mean solar time, or clock time.
The Coordinated Universal Time (UTC) standard, the primary time standard by which the world regulates clocks and time, allows leap seconds to be applied at the end of any UTC month. The first preference is given to June and December and the second to March and September.
Leap seconds are announced only six months in advance—typically in January and July. Till date, all of them have been inserted either at the end of 30 June or 31 December. The last one was announced on 30 June 2012.
The 26th adjustment will be done on 30 June 2015.
Given that the world is becoming increasingly digital with over 3 billion Internet users and close to 2 billion smartphone users (many of whom use the global positioning system or GPS), leap seconds—in this case a positive one that will add one second to atomic clocks (all leap seconds have been positive since 1972)—will matter all the more.
In the past, leap seconds have caused outages at popular websites such as Reddit, Yelp and Foursquare. A 2012 leap second even led to 400 flight delays in Australia, forcing airport workers to conduct check-ins by hand rather than using computers, according to a 2 July 2012 article in The Guardian.
“The failure in Amazon’s cloud systems affected a wide range of companies, including the photo-sharing service Instagram, the “pinning" service Pinterest, and Netflix," the report had said.
According to an 18 May article in The Wall Street Journal, the 30 June “leap second is the first to take place during active trading hours since 1997", when there was no algorithmic trading.
The report added that while US stock exchanges including the New York Stock Exchange’s NYSE Arca, Nasdaq OMX Group Inc.’s Nasdaq and BATS Global Markets “will close their after-hours sessions 30 minutes before their regular 8pm close", European markets including London “are largely closed when the event is scheduled to occur".
Meanwhile, the Australian Securities and Investment Commission (ASIC) has already issued an advisory on its website, recommending that “financial markets stakeholders review their information systems and processes and address any potential business impacts.
“Relevant questions to ask include: Will the insertion of the leap second impact time-stamps on your trading system? Do you have any critical processes that are initiated at 10.00.00 on 1 July 2015? Will you be sourcing any time critical reference information at the relevant time and, if so, how will this be impacted by the leap second change?"
The Australian Securities Exchange (ASX) has responded by stating on its website that the “...net effect of the insertion of the Leap Second is that, depending on your firm’s time source and your system’s handling of timestamps, international time may appear to go back one second on some systems thus potentially impacting system stability. ASX’s approach to the Leap Second will use the built-in functionality of the Precision Time Protocol (PTP) servers to deliver a time sequence from 9:59:59 to re-align with UTC..."
The problem is that not all clocks implement leap seconds similar to the UTC.
Leap seconds in Unix (on systems that run the operating system called Unix) time are typically implemented by repeating the last second of the day. Network Time Protocol (NTP), on the other hand, freezes time during the leap second. It is a networking protocol for clock synchronization between computer systems over packet-switched data networks.
NTP, one of the oldest Internet protocols in current use, was originally designed by David L. Mills of the University of Delaware, who still oversees its development. It uses a modified version of Marzullo’s algorithm to select accurate time servers and is designed to mitigate the effects of variable network latency. The algorithm was invented by Keith Marzullo, professor in the Department of Computer Science and Engineering at University of California, San Diego.
But what will be the effect on GPS, which is increasingly being used to time-stamp business transactions, in data and wireless networks, in communication systems, power grids, and to synchronise reporting of weather from doppler radars?
Fortunately, GPS satellites contain multiple atomic clocks that contribute very precise time data to the GPS signals.
GPS receivers decode these signals, effectively synchronizing each receiver to the atomic clocks. However, on 26 May, the US Civil GPS Service Interface Committee (CGSIC) issued a notice, recommending that receivers be built in accordance with GPS interface specifications (the latest specification may be found here) so that they “automatically adjust UTC time of day outputs to reflect leap second changes".
In a “Best practices" note on its website, the agency also recommends that if you know your system will not work through the leap second on 30 June 2015, you may “...isolate the system from any external timing sources and manually insert the additional second at 23:59:59 UTC".
Google Inc says it tackles the issue with its “leap smear" concept, according to a 15 September, 2011, blog on its official site.
According to the note by Christopher Pascoe, a Site Reliability Engineer from the company’s Site Reliability team—responsible for keeping Google’s services and data centers up and running continuously—Google modified its internal NTP servers to gradually add a couple of milliseconds to every update, varying over a time window before the moment when the leap second actually happens.
“This meant that when it became time to add an extra second at midnight, our clocks had already taken this into account, by skewing the time over the course of the day. All of our servers were then able to continue as normal with the new year, blissfully unaware that a leap second had just occurred," Pascoe explained in the note.
Google plans to use the “leap smear" technique this time too
Unlock a world of Benefits! From insightful newsletters to real-time stock tracking, breaking news and a personalized newsfeed – it's all here, just a click away! Login Now!
