Art of Codeartofcode
Lesson 6 of 17

16. Consistency and CAP

1
The picture to keep in your head
2
16.1 What does "consistent" actually mean?
3
16.2 What is strong consistency?
4
16.3 What is eventual consistency?
5
16.4 What are read-your-writes and monotonic reads?
6
16.5 What is a network partition?
7
16.6 What does the CAP theorem actually say?
8
16.7 What do CP and AP systems look like in practice?
9
16.8 What is PACELC?
10
16.9 How do quorums give tunable consistency?
11
16.10 How do you choose consistency for each feature?
12
Memory card and where the analogy breaks
13
Practice: five levels
14
Interview questions
15
Solutions
16
Sources
17
Code lab
16. Consistency and CAP

16.5 What is a network partition?

3 min read+10 XP

CAP and beyond

At the bank: The phone line between the counters is cut. Each counter is fine, customers are still coming in, but they can't reach each other. Worse: from counter 2's side, "the line is cut" and "counter 1 is just slow to pick up" look exactly the same.

In the system: A network partition is when some machines can't communicate with others, even though each is still running. Causes: a failed switch or cable, a misconfigured firewall, a cloud provider's network problem between zones or regions, or a machine so overloaded it can't respond in time.

The key insight is that you can't tell a slow node from a dead one or an unreachable one. A request that hasn't been answered after 2 seconds might arrive in 3 seconds, or never. That's why systems use timeouts and must decide what to do when one fires.

Partitions are not optional. Any system with more than one machine talking over a network will, sooner or later, have messages lost or delayed. You don't get to choose whether partitions happen, only how your system behaves when they do.

Memory code: "A cut line and a slow line look the same from your side."