liveBook · Manning

Parallelism

As explained in Fast by Default

Parallelism is the simultaneous execution of multiple independent units of computation or work.

Parallelism runs independent work at the same time instead of extending a user-visible sequence step by step. It reduces latency when operations have no true dependency, but requires coordination and error handling. section 4.1 section 5.2

Use parallelism where tracing shows avoidable serialization:

  • Map dependencies first; preserve ordering where one operation genuinely needs another. section 4.1
  • Parallelize independent requests, such as fetching user data, preferences, and content that do not depend on one another. section 4.1
  • Keep nonessential work off the critical path by deferring analytics, secondary data, or other operations that do not affect the initial view. section 4.1
  • Coordinate asynchronous results deliberately; parallel execution improves progress but adds complexity around completion and failures. section 5.2

Aggregating service responses

A backend-for-frontend or facade can call multiple services in parallel, combine their responses, cache results, and return only the fields the interface needs. This improves the client experience without changing legacy or third-party services. section 5.5.1

Measure before changing the sequence. Use traces and platform profilers to identify what blocks the user, then compare the critical path before and after parallelization. A shorter total workload is not required for the interface to feel faster: delivering independent results incrementally can replace blank waiting with visible progress. section 4.1 section 5.2

How Fast by Default covers it

How deep it goes
Explains dependency mapping, deferred work, and asynchronous trade-offs, but stops short of language-specific implementation details.
Examples
Uses service aggregation, carrier-rate checks, independent page requests, and mobile startup data as concrete examples.
What it assumes
Assumes basic operations and dependencies, while teaching how to identify true dependencies and distinguish essential from deferrable work.
What you can do after
Diagnose avoidable serialization with traces, parallelize independent work, and compare its effect on the user-visible critical path.
How it connects
Connects parallelism with critical path, asynchronous execution, caching, lazy loading, progressive delivery, and facade pattern.

Also explained in

Compare how 55 books explain parallelism →

Related concepts

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