Distributed Systems Fallacies Pocket Reference

In 1994, Peter Deutsch wrote down the seven assumptions every engineer new to distributed systems made — and that every distributed system eventually punished them for making. James Gosling added an eighth. Thirty years on, the fallacies have not been outgrown; they have been hidden, by an industry that got very good at presenting the network as something it has never been: reliable, fast, secure, free, homogeneous, and managed by one benevolent administrator. This pocket reference walks the list one chapter at a time — what each fallacy means in production, how cloud abstractions disguise it, and the engineering it takes to build systems that survive when the lie breaks.

Table of Contents

1. 0 Origins
2. 1 The Network Is Reliable
3. 2 Latency Is Zero
4. 3 Bandwidth Is Infinite
5. 4 The Network Is Secure
6. 5 Topology Doesn't Change
7. 6 There Is One Administrator
8. 7 Transport Cost Is Zero
9. 8 The Network Is Homogeneous
10. 9 The Meta-Fallacy