Time feels simple on one machine. In a distributed system, it is not simple at all.

Every machine has its own clock. Those clocks slowly drift apart. The software that tries to fix them also has to talk over the network, and the network is never perfectly predictable.

This matters because engineers often use timestamps to make important decisions:

Wall-clock time is useful, but it is not safe when an important rule depends on exact ordering across machines.

This chapter covers why clocks disagree, when to use wall clocks or monotonic clocks, what NTP and PTP can do, where clock skew causes bugs, and when physical time is good enough.