Introduction

Functions at the edge, global replication, and pay-per-request are becoming one offering from major clouds and CDN providers. The result is a convergence: you design for “run everywhere, scale to zero,” and the platform handles placement and scaling. We map the landscape and common patterns.

What “Convergence” Means

Single abstraction, many regions
You deploy one function or worker; it runs at the edge or in regional backends depending on config and traffic. We explain how providers route and replicate so you can reason about latency and cost.

State and storage
Edge and serverless are stateless by default. We cover KV, Durable Objects, and DB proxies that let you build stateful apps without managing servers.

Use Cases and Gotchas

APIs, auth, and personalization
Low-latency APIs, token validation, and A/B logic are natural fits. We outline an example and discuss cold starts and limits.

When not to converge
Long-running jobs, heavy compute, and strict data residency may still need traditional or sovereign setups. We help you decide.