Posted by Richard N Williams on March 25th, 2009
With a variety of acronyms and timescales the world of time synchronisation can be quite confusing here are some frequently asked questions we hope will help enlighten you.
What is NTP?
NTP is a protocol designed to synchronize computer networks across the internet or LAN (Local Area Networks). It is not the only time synchronization protocol available but it is the most widely used and the oldest having been conceived in the late 1980’s.
What are UTC and GMT?
UTC or Coordinated Universal Time is a global timescale, it is controlled by highly accurate atomic clocks but kept the same as GMT (Greenwich Meantime) by the use of leap seconds, added when the Earth’s rotation slows down. Strictly speaking GMT is the old civil timescale and based on when the sun is above the meridian line, however, as the two systems are identical in time thanks to leap seconds, UTC is often referred to as GMT and vice versa.
And a NTP Time Server?
These are devices that synchronize a computer network to UTC by receiving a time signal and distributing it with the protocol NTP which ensures all devices are running accurately to the timing reference.
Where to get UTC time from?
There are two secure methods of receiving UTC. The first is to utilize the long wave time signals broadcast by NIST (WWVB) NPL in the UK (MSF) and the German NPL (DCF) The other method is to use a the GPS network. GPS satellites broadcast an atomic clock signal that can be utilised and converted to UTC by the GPS NTP server.
Posted by Stuart on March 24th, 2009
The NTP GPS server is a dedicated device that uses the time signal from the GPS (Global Positioning System) network. GPS is now a common tool for motorists with satellite navigation devices fitted to most new cars. But GPS is far more than just an aid for positioning, at the very heart of the GPS network is the atomic clocks that are inside each GPS satellite.
The GPS system works by transmitting the time from these clocks along with the position and velocity of the satellite. A satellite navigation receiver will work out when it receives this time how long it took to arrive and therefore how far the signal travelled. Using three or more of these signals the satellite navigation device can work out exactly where it is.
GPS can only do this because of the atomic clocks that it uses to transmit the time signals. These time signals travel, like all radio signals, at the speed of light so an inaccuracy of just 1 millisecond (1/1000 of a second) could result in the satellite navigation being nearly 300 kilometres out.
Because these clocks have to be so accurate, they make an ideal source of time for a NTP time server. NTP (Network Time Protocol) is the software that distributes the time from the time server to the network. GPS time and UTC (Coordinated Universal Time) the civil timescale is not quite the same thing but are base don the same timescale so NTP has no trouble converting it. Using a dedicated NTP GPS server a network can be realistically synchronised to within a few milliseconds of UTC
The GPS clock is another term often given to a GPS time server. The GPS network consists of 21 active satellites (and a few spare) 10,000 miles in orbit above the Earth and each satellite circles the Earth twice a day. Designed for satellite navigation, A GPS receiver needs at least three satellites to maintain a position. However, in the case of a GPS clock just one satellite is required making it far easier to obtain a reliable signal.
Each satellite continuously transmits its own position and a time code. The time code is generated by an onboard atomic clock and is highly accurate, it has to be as this information is used by the GPS receiver to triangulate a position and if it was just half a second out the Sat Nav unit would be inaccurate by thousands of miles.
Posted by Richard N Williams on March 20th, 2009
Most people have vaguely heard of the atomic clock and presume they know what one is but very few people know just how important atomic clocks are for the running of our day to day lives in the twenty first century.
There are so many technologies that are reliant on atomic clocks and without many of the tasks we take for granted would be impossible. Air traffic control, satellite navigation and internet trading are just a few of the applications that are reliant on the ultra precise chronometry of an atomic clock.
Exactly what an atomic clock is, is often misunderstood. In simple terms an atomic clock is a device that uses the oscillations of atoms at different energy states to count ticks between seconds. Currently caesium is the preferred atom because it has over 9 billion ticks every second and because these oscillations never change it makes them a highly accurate method of keeping time.
Atomic clocks despite what many people claim are only ever found in large scale physics laboratories such as NPL (UK National Physical Laboratory) and NIST (US National Institute of Standards and Time). Often people suggest they have an atomic clock that controls their computer network or that they have an atomic clock on their wall. This is not true and what people are referring to is that they have a clock or time server that receives the time from an atomic clock.
Devices like the NTP time server often receive atomic clock signals form places such as NIST or NPL via long wave radio. Another method for receiving time from atomic clocks is using the GPS network (Global Positioning System).
The GPS network and satellite navigation are in fact a good example of why atomic clock synchonization is much needed with such high level of accuracy. Modern atomic clocks such as those found at NIST, NPL and inside orbiting GPS satellites are accurate to within a second every 100 million years or so. This accuracy is crucial when you examine how something like a cars GPS satellite navigation system works.
A GPS system works by triangulating the time signals sent from three or more separate GPS satellites and their onboard atomic clocks. Because these signals travel at the speed of light (nearly 100,000km a second) an inaccuracy of even one whole millisecond could put the navigational information out by 100 kilometres.
This high level of accuracy is also required for technologies such as air traffic control ensuring our crowded skies remain safe and is even critical for many Internet transactions such as trading in derivatives where the value can rise and fall every second.
Posted by Richard N Williams on March 16th, 2009
If you are reading this then you are probably aware of the importance time plays in IT systems and computer networks. Most computer administrators appreciate that precise time and accurate synchronisation are an important aspect of keeping a computer network error free and secure.
And yet, despite its importance many network administrators still rely on the Internet as a source of UTC time for their networks (UTC – Coordinated Universal Time), primarily because they see it as a quick and more importantly a free method of time synchronisation.
However, the drawbacks in using these free services may cost a lot more than the money saved on a dedicated NTP time server.
NTP (Network Time Protocol) is now present on nearly all computers and it is NTP that is used to synchronise computer systems. However, if an Internet time source is used then the source is outside the network firewall and this creates a serious vulnerability. Any external time source will require a port to be left open in the firewall to allow the time information packets through and this opening is too easy a way to exploit a network which can become victim to a DDOS attack (Distributed Denial of Service) or even allow malicious programmes through to take control of the machines themselves.
Another problem is the availability of stratum 1 time sources across the internet. Most online time sources come from stratum 2 time servers. These are devices that receive the time from a time server (stratum 1) that originally gets the information from an atomic clock (stratum 0). While stratum 2 devices can be just as accurate as stratum 1 time servers, across the internet without NTP authentication the actual accuracy can not be guaranteed.
Furthermore, internet time sources have never been considered accurate or precise with surveys showing over half being inaccurate by over a second and the rest dependent on the distance from client as to whether they can provide any useful accuracy. Even organisations such as NIST publish advisory notices on their time server pages about it unable to guarantee security or accuracy and yet millions of networks are still receiving time from across the internet.
With the decline in cost of dedicated radio referenced NTP time servers or GPS NTP server there has never been a better time to get one. And when you consider the cost of a computer breach or crashed network the NTP server will have paid for itself many times over.
Posted by Richard N Williams on March 11th, 2009
Radio signal goes dead for several hours
The long wave transmissions such as MSF (NPL) or WWVB (NIST) are broadcast from large antennas that often need maintenance. This often requires a shut down of the broadcast while it is being done. These outages are normally posted with at least three months notice on the websites of the signals controllers (and can be automatically emailed if you register) to give prior notice.
These outages only tend to last a few hours leaving your computer network reliant on the electronic system clocks but it is doubtful there will be too much drift in that time (and any drift will be accounted for once the signal is back on. If these outages could be a potential problem than a simple solution is to invest in a dual system that will receive both GPS time server and radio signals ensuring a continuous time signal.
No time signal coming in despite the time server being powered up
This is most often caused by either lack of power going to the antenna or failing to connect to site the antenna where it can have a clear view of the sky. GPS antennas may have battery or power connections so it is always worth checking before switching the device on. Ensuring the antenna can ‘view’ the satellites when using GPS time servers is also important, remembering that windows and skylights may prevent signals getting through.
When using radio time reference such as MSF, DCF or WWVB the NTP server antennas can receive the long wave signal indoors but they are vulnerable to topography and local interference. If there is no signal or only a weak signal then try moving the antenna around until the signal strength increases enough.
Often users of these time and frequency signals find that the signal is weak throughout the day but is boosted at night. This is because the signals are ground state but have a residual skywave which can bounce of the ionosphere during the coolness of the night (ionospheric propagation).
Some users of these signals may find that despite being well within range the local topography can prevent a strong enough signal from getting through.
Posted by Richard N Williams on March 9th, 2009
NTP servers are the easiest, most accurate and secure method of receiving a UTC time source (Coordinated Universal Time). Most dedicated NTP time servers will run in the background automatically synchronising the devices on a network completely automatically.
However, there are some common problems that occasionally occur in using a network time server but fortunately most can be solved relatively easily.
Losing A GPS time signal
GPS is one of the most efficient sources of UTC time. The GPS signal is available literally anywhere on the planet where there is a clear view of the sky. At any one time there are at least three satellites within range of any location and unlike radio referenced transmissions there are no maintenance outages so the signal is always uninterrupted.
However, some people find that they keep losing their GPS signal when using a GPS NTP time server. Very rarely this can be caused by extra terrestrial occurrences (solar flares – not little green men), however more commonly signal loss occurs when there has been insufficient time give for the initial acquisition lock.
To ensure a continuous signal make sure you follow manufacturer’s recommendation for obtaining acquisition. This usually means leaving the GPS time server to get a good lock for at least 24 hours (so all satellites have been in view). If not enough time is given to this then it is possible the GPS time server will lose a satellite and therefore timing information.
One second delay in a radio clock compared to internet or GPS
This is a very frequent occurrence when using a radio time server using signals such as the MSF transmission broadcast by the UK’s National Physical Laboratory. This occurs normally after the insertion of a Leap Second. Leap seconds are introduced once or twice a year to compensate for the slowing of the Earth’s rotation and to keep UTC in line with the Greenwich Meridian.
While NTP will automatically account for leap seconds with signals like the MSF it can often take some time as there is no Leap Second announcement. This announcement normally allows NTP to prepare for the leap second (which normally occurs in the last second of the last day in June or December). As signals such as MSF do not announce the upcoming leap second it can take some time for it to be accounted for. In some cases it can take a few days in others minutes. A simple solution is to manually announce the leap second.
However, if this is not done, NTP will eventually discover the leap second and adjust the network clocks.
Posted by Stuart on March 6th, 2009
A network time server (commonly referred to as the NTP time server after the protocol used in synchronisation – Network Time Protocol) is a device that receives a single time signal and distributes it to all devices on a network.
Network time servers are preferred as a synchronisation tool rather than the much simpler internet time servers because they are far more secure. Using the internet as a basis for time information would mean using a source outside the firewall which could allow malicious users to take advantage.
Network time servers on the other hand work inside the firewall by receiving source of UTC time (Coordinated Universal Time) from either the GPS network or specialist radio transmissions broadcast from national physics laboratories.
Both of these signals are incredibly accurate and secure with both methods providing millisecond accuracy to UTC. However, there are downsides to both systems. The radio signals broadcast by nation time and frequency laboratories are susceptible to interference and locality, while the GPS signal, although available literally everywhere on the globe can occasional be lost too (often due to bad weather interfering with the line-of-sight GPS signals.
For computer networks where high levels of accuracy are imperative, dual systems are often incorporated. These network time servers receive the time signal from both the GPS network and the radio transmissions and select an average for even more accuracy. However, the real advantage of using a dual system is that if one signal fails, for what ever the reason, the network will not have to rely on the inaccurate system clocks as the other method of receiving UTC time should still be operational.
Posted by Richard N Williams on March 4th, 2009
Atomic clocks have been around since the 1950’s when NPL (National Physical Laboratory) in the UK developed the first reliable caesium based clock. Before atomic clocks, electronic clocks were the most accurate method of keeping track of time but while an electrical clock may lose a second in every week or so, a modern atomic clock will not lose a single second in hundreds of millions of years.
Atomic clocks are not just used to keep track of time. The atomic clock is an integral part of the GPS system (Global Positioning System) as each GPs satellite has its own onboard atomic clock that generates a time signal that is picked up by GPS receivers who can calculate their position by using the precise signal from three or more satellites.
Atomic clocks need to be used as the signal s from the satellites travel at the speed of light and as light travels nearly 300,000 km each second any slight inaccuracy could put navigation out by miles.
A GPS time server is a network time server that uses the time signal from the GPS network’s satellites to synchronise the time on computer networks. A GPS time server often uses NTP (Network Time Protocol) as a method of distributing time which is why these devices are often referred to as NTP GPS time servers.
Computer networks that are synchronised using a dedicated time server are normally synchronised to UTC (Coordinated Universal Time) and while the GPS signal is not UTC, GPS time, like UTC, is based on International Atomic Time (TAI) and is easily converted by NTP.
Posted by Stuart on March 2nd, 2009
A time server is a crucial piece of kit for any network. Time synchronisation is imperative in keeping a network secure and reliable. Time synchronisation, however, need not be the headache many administrators assume it is going to be.
Most of the difficulties of time synchronisation have been taken care of thanks to the protocol NTP (Network Time Protocol). Whilst NTP is not the only time synchronisation software available it is by far the most widely used (due mainly to the fact that it has been around since the 1980’s and is sill being developed today).
NTP uses a single time source and distributes it from machine-to-machine checking each PC or device for drift then adjusting for it. NTP is normally installed on Windows and Linux systems (or at least a simplified version called SNTP) although it is freely downloadable from the NTP homepage. While NTP can quite easily receive any time source from the Internet this can cause major security issues no to mention a lack of accuracy that many online NTP servers suffer from.
The most accurate and secure method is to use an external network time server as these sit within the firewall. They are also receive a UTC (Coordinated Universal Time) reference direct from an atomic clock which makes them stratum 1 devices. Most internet time servers are stratum 2 servers. NTP uses strata to define how far away a server is from the source so an atomic clock is a stratum 0 device while a computer that receives time direct from a NTP server becomes a stratum 2 device and so on.
The only decision that really needs to be made when installing a dedicated NTP time server is which time reference is best. There are two main methods of receiving a secure, accurate and authenticated UTC time reference; the GPS network (Global Positioning System) or national physics laboratories long wave radio transmissions.
The latter system is not available in every country although the USA, UK and Germany have strong signals known as WWVB, MSF and DCF respectively. These can often be picked up outside the borders of these countries although the signals are vulnerable to interference, outages and local topography.
A GPS NTP server system is less vulnerable to these things and as long as there is a clear view of the sky (such as a rooftop or open window) the GPS time signal can be picked up anywhere on the globe.
Posted by Richard N Williams on February 25th, 2009
NTP homepage– The home for the NTP Project who provides support and additional development resources for the Official Reference Implementation of NTP.
NTP Project support pages
THE NTP pool – list of public servers
NPL – The National Physical Laboratory in the UK who control the MSF radio signal.
The University of Delaware and David Mills’ information page, Professor Mills is the original inventor and developer of NTP
David Mills’ list of Public NTP Time Servers a list of public NTP servers
National Institute of Standards and Technology (NIST) who operate the USA’s WWVB radio signal
Europe’s largest supplier of NTP server related products.
Galleon UK – NTP server products for the UK
NTP Time Server .com – one of the largest time and frequency suppliers in the United States
NTP – Wikipedia article on NTP
NTP server checker – free tool to ensure time server accuracy