liveBook · Manning

Fallacies of Distributed Computing

As explained in Grokking Software Architecture

The eight commonly cited false assumptions about networks and distributed environments, such as assuming that networks are reliable or have zero latency.

The Fallacies of Distributed Computing are eight false assumptions about networks and distributed environments. They are a useful checklist when considering microservices: splitting work across services means network costs and failures become part of the design. section 12.6.1

The eight assumptions are:

The latency tax on a dashboard

A dashboard needs a profile, recent orders, and shipping status. In one monolith scenario, a SQL query with joins takes roughly 5 ms. A microservice gateway instead makes three HTTP calls: 35 ms for the user service, 40 ms for orders, and 35 ms for shipping, including serialization. If made sequentially, those calls add up to 110 ms. A network trip that looks small on its own can become noticeable when a request needs several of them. section 12.6.1

Before treating a service boundary as free, ask which assumptions it depends on and what costs or failures follow when they do not hold. In particular, count network trips when comparing a distributed design with a monolith. section 12.6.1 section 12.10

How Grokking Software Architecture covers it

How deep it goes
Lists all eight assumptions and works through network and serialization costs in a dashboard example.
Examples
Compares a dashboard's joined SQL query with HTTP calls to user, order, and shipping services.
What you can do after
Evaluate distributed designs by identifying false network assumptions and their potential costs.
How it connects
Connects microservices and monoliths through network latency and serialization costs.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Grokking Software Architecture. The section links point to the parts of the book it draws on.