Part 1 of 5 · 1 min read
Why this page comes first
In this module you will
- Explain why a transaction spanning multiple nodes needs more coordination than a single-node transaction.
- Walk through the two phases of two-phase commit, and what each phase guarantees.
- Explain why the coordinator is a single point of failure in two-phase commit.
- Judge whether a described multi-node operation actually needs two-phase commit's guarantees.
The previous page covered the general unreliability of networks, clocks, and partial failures. This page covers a concrete problem that unreliability makes hard. A single transaction may touch data on more than one node. It must either commit everywhere or abort everywhere, never partly one and partly the other.