Hacker News Reader

Comment by prab97 | Hacker News Reader

That reminds me of a very important aspect of distributed systems (computer science) - there is no global clock!

UTC is a global clock. If I show the time in Sydney and the time here in UK, Sydney will be 150ms behind to me, and 150ms ahead to a viewer in Sydney, but I know the distance therefore I know if the clocks are in sync.

The time dilation between the two places is on the order of femtoseconds/second, a millionth of a clock cycle of a cpu.

UTC is a local clock on which other local clocks are synchronized.

Sounds to me like you could have equally started that post with "UTC is not a global clock".

It's a global clock. UTC is the same time in Miami as it is in Singapore.

The accuracy of the delivery of that timesource is 10ns in theory, and upto 1000ns in practice.

CET and EST are also the same time in Miami as in Singapore. That's just how time zones work. The distinction with UTC is that it isn't tied to any particular physical location, it's not the "time zone" for anywhere.

But UTC is ultimately defined by consensus. We need a reference clock and we need to be able to measure or estimate our skew in order to sync to it.

How do you know you have the correct (to the accepted precision) time?

GPS provides the correct time anywhere on the planet, it's a universal (as far as earth goes) clock, accurate to 10 nanoseconds (3m). The different reference planes that opposite sides of the planet (at the equator) gives you a precision of femtoseconds (micrometers at light speed) so makes no difference there.

The quality of GPS receiver of course is important, it may reduce your precision to to 100ns or even 1000ns, but it will give you the right time. Someone equidistant to two GPS synced clocks will see them both at the same time (within a microsecond)

Reply on news.ycombinator.com