Concurrency and Async

Verified status as of March 28, 2026.

Quick View

• Complexity: Intermediate
• Typical time: 10-15 minutes
• Outcome: practical mental model for concurrency by runtime in FastFN

Runtime Concurrency Model

Runtime	Model	Practical note
Node.js	event loop + async IO	avoid blocking CPU work in request path
Python	process + worker model	choose async libraries for IO-heavy tasks
Rust	compiled handler with explicit async model choices	keep serialization and IO bounded
PHP	persistent daemon handling invocation cycles	minimize per-request bootstrap cost
Go	goroutines + concurrency primitives	keep shared state explicit and race-safe
Lua	lightweight runtime inside OpenResty flow	keep handler logic short and IO-aware

What to Optimize First

1. external IO latency (DB/APIs)
2. payload size and serialization cost
3. timeout budgets and retries

Validation

• p95 latency is stable under expected concurrency.
• timeouts are explicit and tested.
• retries are bounded and idempotent.

Troubleshooting

• If p95 degrades, profile IO before optimizing handler CPU.
• If timeouts spike, check downstream rate limits and pool sizing.

Related links

• Performance benchmarks
• Run and test
• Platform runtime plumbing

Last reviewed: March 28, 2026 · Docs on fastfn.dev