Question

If a distributed system must choose between being consistent and being available during a network partition, which one must it sacrifice to strictly follow the CAP theorem?

Accepted Answer

The CAP theorem says you can only pick two things out of three for a computer system. The three things are consistency, availability, and partition tolerance. Consistency means every person looking at the system sees the exact same data at the same time. Availability means the system always answers your request, even if it is slow or the data is a bit old. A network partition is when the computers in a system stop being able to talk to each other because a cable breaks or a wire goes bad. When this happens, you have to choose between being consistent or being available. You cannot have both. If you choose consistency, you must stop answering requests while the computers are broken because you cannot guarantee the answer is the most recent one. This means you sacrifice availability. If you choose availability, you keep answering requests even though you know the information might be outdated or wrong. This means you sacrifice consistency. Think of it like a group of friends trying to decide on a movie while their phones are broken. If they insist on everyone agreeing on the exact same movie at the same time, they must stop talking until the phones are fixed, which means they are not available to make a choice. If they just pick a movie quickly without checking with everyone, they are available, but they lose consistency because someone might end up at the wrong theater.

Home → All Courses → Engineering and Technology Courses → Distributed Systems & IoT Architecture → Flashcard

If a distributed system must choose between being consistent and being available during a network partition, which one must it sacrifice to strictly follow the CAP theorem?

Community Answers