00:03
How might the clock in two computers that are linked by a local network be synchronized without preference to external time sources and also what factor limit the accuracy of the procedure you have described? so synchronized clock of two computers on a local network, which has no external time resource basically is done by a message exchange.
00:37
So basically, basically, the message exchange will estimate offset and network delay between them.
00:44
So say we call a pretty common method in our case simplify the version of the network time protocol ntp.
00:58
So network time protocol and it basically like this.
01:03
So say you have computer a we call the client and it sends a message to b.
01:22
B is what we called a server.
01:26
So this is at time 1 according to 8 o 'clock.
01:50
Now, b received at t2 according to b's clock.
01:55
Immediately send a reply back with time step t2.
02:44
So then a received the time at t3.
03:08
So again, this is back to a's clock.
03:13
So we have three different times.
03:15
So basically, t1, t2, and t3.
03:18
And so from there, you can see that adjust the estimate, offset can be done by t2 minus t1 plus t3 divided by 2.
03:43
So then a can adjust the clock by its estimated offset and then synchronize with b.
03:50
So basically you have a round -trip delay between t3 and t -1 and t3.
03:57
So then the a can adjust the clock by the off -s set.
04:04
Now the next step is what is the question asked about what factor limit the accuracy of the procedure you have described.
04:22
So the accuracy limit variable network delay is asymmetric path, clock drift, and processing delay.
05:14
So for example, when we talk about the variable network delay, that means the message may not always take the same time to travel...