Edge Computing
Running application logic on servers physically close to users — at CDN points of presence — rather than in one central data centre.
· Reviewed by senior engineers
Edge computing means running application code at the network edge — in the CDN's hundreds of points of presence around the world — rather than in one central data centre or cloud region. Cloudflare Workers, Vercel Edge Functions, Fastly Compute@Edge and AWS Lambda@Edge are the leading platforms.
The motivation is latency. If the user is 5 ms from the nearest edge node and 200 ms from your origin, doing personalisation, A/B testing, auth checks or geo-routing at the edge gives you sub-50 ms total response times. Edge is also where bot detection, rate limiting and security rules belong because every request passes through.
The constraints are real. Edge runtimes are usually V8 isolates, not full Node.js — no native modules, tight CPU and memory limits, no direct database connections. State has to live elsewhere (KV stores, durable objects, regional databases) and you have to think carefully about where data sits relative to the request.
We use edge functions for the things they are good at — auth gating, geo-routing, header rewriting, lightweight personalisation — and keep heavy compute and stateful logic in regional or serverless tiers behind them. Used well, edge functions are the single biggest performance lever available on the modern web.