Bookmark/Search this post with:

                                                 

…it’s getting better all the time… A most fascinating insight into the history of NTP

Some things deteriorate with age - your great Aunt Wilma for example. Other things, such a full bodied bottle of Merlot, well they blossom, become more sophisticated and slightly sexier. Like the bottle of Merlot collecting dust in your basement, NTP (Network Time Protocol) is also something that is certainly getting better with time.

‘But what is NTP?!’ I hear you cry…

Well, NTP (yes, that’s right – Network Time Protocol) is used to synchronize the clocks of computer systems. It uses a trusted time reference to ensure accurate time-keeping. It is powerful enough to synchronize network clocks with millisecond accuracy – and this accuracy is just going to improve as NTP continues to evolve.

‘But why are you telling me this? What is all the fuss about?’

Without NTP, time would not be consistent over any two computers, let alone a network of hundreds, or thousands. Seconds would be lost here and more seconds would be gained there. It would cause major confusion.

Don’t believe me? Remember the panic and paranoia that hit the headlines with regards the Millennium Bug?! Yup, we were all sure that on midnight when the year ticked over to 2000 the world was surely, most definitely going to end. But guess what? It didn’t! But the fear was not completely unjustified…. Time needs to be carefully coordinated, reliable and accurate. If it is not, then it might cause DNS caches to expire, causing such horrors to happen such as the Millennium Bug. Thankfully, due to the complex, and frankly brilliant, nature of NTP, the end of the world did not occur.

We also need time to be spot on if we’re to avoid crashes when we’re flying back from our hols in Spain, if we want to buy something off an auction website and also to schedule TV networks. Without our computers and internet servers running on accurate time, the world would be full of all sorts of confusion. Data files might expire before they were even created and you might receive an email before the sender had even written it. Crazy.

Now I have your attention, and you are definitely convinced that our lives would be a far worse place without the wonders of NTP, you are probably scratching your head as to think – where on earth did NTP come from?

Here’s one of the clever chappies who helped develop the idea, Professor David L. Mills, from the University of Delaware.

Not only does he have the ability to grow impressive facial hair, but during the 80’s he was all about the synchronizing of computer network time. The protocols that he developed have evolved into the Network Time Protocol.

During the early 1980’s, an Internet Clock Service was developed. It came about as there was need for clock synchronization for the ‘HELLO routing protocol’ (one of the first examples of messages being sent electronically over a very early system of what we now know as the ‘can’t-live-without-it-internet’.) The service was based on very basic calculations and only had an accuracy of several hundred milliseconds. Still, it was a start…

In 1988, the specification of the protocol and accompanying algorithms (that’s complicated calculations for the non technical talkers of us…) were completed for the first version of NTP. As well as the specification of the protocol and algorithms, it also included early versions of other aspects of NTP such as the clock filter (which are used to cut out computer jitter).

1989 saw the creation of NTP Version 2. The second version included features such as NTP Control Message Protocol for use in managing NTP servers and clients and the cryptographic authentication scheme based on symmetric-key cryptography. Both of these have survived to the present day version of NTP.

However, 1989 also saw another time synchronization protocol come on the scene - Digital Time Synchronization Service (DTSS). Both the DTSS and the NTP communities had much the same goals – that of accurate and reliable time – but had different strategies for achieving them. The people on the side of NTP thought that DTSS had a problem with losing accuracy as DTSS did not discipline the clock frequency. On the side, the DTSS community criticised how NTP lacked the formal correctness principles in the design process.

Rather than getting mad, the people involved with NTP did what anybody else would do in their situation. They ‘borrowed’ the ideas from DTSS and included them with their own. In 1992, thanks to combining the good ideas of DTSS and NTP, the official internet standard of NTP (version 3) was created. NTP Version 3 introduced formal correctness principles, revised algorithms and a broadcast mode was added to the protocol.

In 1994, a kernel model for precision timekeeping described a new implementation and interface. The implementation could keep time with a precision of up to one microsecond. Since then, both the specification and the implementation of NTP is being continuously improved with version 4 underway. Version 4 includes such shiny, spangly, must have features regarding automatic configuration, reliability, Internet traffic reduction and authentication. A new kernel clock model is also being developed which can keep time with a precision of up to one nanosecond.

So, I think you’ll agree that the development of NTP is surely a marvelous one. As time goes on, so does the improvement of NTP and it is only going to get better and better with further improvements. And just as well, for where would we be without it? Probably, stuck in yesterday’s time zone reading tomorrow’s emails…

Author Biography
Claire Gamble is a technical author and works on the development of time server solutions that synchronize time on computers and computer networks. Click here to find out more about NTP Server and GPS time server solutions.