Part 1 of 6 · 1 min read
Why this page opens the second half of the level
In this module you will
- Explain partial failure and why it makes distributed systems hard.
- List the reasons a request can get no reply, and say why a timeout cannot tell them apart.
- Choose a timeout with the trade-off in mind and say why queues cause variable delay.
- Tell a time-of-day clock from a monotonic clock, and say why clocks cannot order events across machines.
- Describe process pauses, how they break leases, and how a fencing token fixes it.
Modules 1 to 3 assumed one database on one machine. From here the level asks what happens when many machines cooperate. The book "Designing Data-Intensive Applications" spends a whole chapter on the answer: nearly everything that can go wrong does, and the program often cannot tell. Later modules (leader election, consistency, consensus) are built on these facts. The numbers are made up to show the idea, unless the text names the book.