Your computer probably does hundreds and thousands of tasks a day. If that is part of a network then the number of tasks could be millions. From sending emails to saving data, and everything else your computer is tasked to do, they are all logged by the computer or server.
Computers use timestamps to logo processes and indeed, timestamps are used as the only method a computer has to indicate when and if a task or application has been conducted. Timestamps are normally a 16 or 32 bit integer (one long number) that counts back the seconds from a prime epoch – normally 01 January 1970.
So for every task you computer does it will be stamped with the number of seconds from 1970 that the transaction was conducted. These timestamps are the only piece of information a computer system has to ascertain what tasks have been completed and what tasks have yet to be instigated.
The problem with computer networks of more than one machine is that the clocks on individual devices are not accurate enough for many modern time sensitive applications. Computer clocks are prone to drift they are typically based on inexpensive crystal oscillator circuits and can often drift by over a second a day.
This may not seem much but in today’s time sensitive world a second can be a long time indeed especially when you take into account the needs of industries like the stock exchange where a second can be the difference in price of several percent or online seat reservation, where a second can make the difference between an available seat and one that is sold.
This drift is also accumulative so within only a few months the computer systems could be over a minute out of sync and this can have dramatic effects on time sensitive transactions and can result in all sorts of unexpected problems from emails not arriving as a computer thinks they have arrived before they have been sent to data not being backed up or lost completely.
A NTP time server or network time server are increasingly becoming crucial pieces of equipment for the modern computer network. They receive an accurate source of time from an atomic clock and distribute it to all devices on the network. As atomic clocks are incredibly accurate (they won’t drift by a second even in a 100,000 years) and the protocol NTP (Network Time Protocol) continually checks the devices time against the master atomic clock time – it means the computer network will be able to run perfectly synchronised with each device within a few milliseconds of the atomic clock.