Part 1 of 7 · 1 min read
Why this page closes the level
In this module you will
- Explain why committing on each node separately breaks atomicity.
- Walk through two-phase commit, name its two points of no return and show why it can block.
- Tell database-internal distributed transactions from heterogeneous ones.
- Design a saga with compensating steps and explain what it gives up.
- Explain the dual write problem and how a log or outbox avoids it.
Module 1 promised atomic transactions on one node. Modules 4 to 8 showed what breaks when many nodes and a network are involved. This page joins them: how to keep a change that touches several machines all-or-nothing, and what to do when the heavy tool for it (two-phase commit) costs too much. The books used are "Designing Data-Intensive Applications", "Understanding Distributed Systems" and "Building Microservices". One section is marked as general knowledge. The numbers are made up to show the idea, unless the text names a book.