Often administrators will boast of their network’s timing accuracy by lamenting that it is all controlled by an atomic clock. Whilst indirectly this may be true what they are referring to is a network time server that receives time from an atomic clock.
Atomic clocks are ridiculously expensive but are by far the most accurate chronometers possible. In fact they are so accurate the international timescale UTC (coordinated universal time) has to have seconds known as ‘leap seconds’ added because the Earth’s rotation is not as precise as our clocks.
Because an atomic clock is so precise they make the ideal sources for UTC time on computer networks; making possible such time sensitive transactions as global trading in stocks and shares, airline reservation and even Internet auctions.
Atomic clocks are known as stratum 0 servers. This is because they sit at the top of the NTP (network time protocol) hierarchical tree. Servers that receive a timing signal; from these stratum 0 atomic clocks are known as stratum 1 devices. It is these stratum 1 devices that are network time servers, used to synchronise a network to.
Machines that use a network time server also have a place on the NTP stratum tree as a stratum 2 server. Computers and devices do not just have to get timing information from a stratum 1 time server but also stratum 2 servers. They can also synchronise with each other.
Network time servers receive the atomic clock timing signal from either a specialist radio transmission in (certain countries only) or from the GPS network. Atomic time signals are available over the internet but these stratum 1 servers are notoriously inaccurate.