What Is Edge Computing?

The Problem with Centralised Servers

A traditional server sitting in us-east-1 adds 150-300ms of network latency for users in Singapore or São Paulo. That latency is irreducible — it is the speed of light across fibre. Edge computing eliminates it by running code in data centres located physically near the user.

Edge vs CDN vs Serverless

• CDN — caches static assets (HTML, CSS, images) at edge nodes; no dynamic computation
• Edge functions — run compute (JavaScript, WASM) at edge nodes; dynamic but within constraints
• Serverless (Lambda, Cloud Run) — full Node.js/Python runtime in a central region; flexible but slower cold starts and higher latency
• Edge — sub-millisecond cold starts, global distribution, V8 isolates (not full OS process)

The Edge Runtime Constraints

Edge functions run in V8 isolates — sandboxed JavaScript environments. They are not full Node.js. Key limitations:

• No Node.js built-ins (no fs, no net, no child_process)
• No long-running connections (WebSockets are supported on Cloudflare; not on all platforms)
• CPU time limits (typically 50ms on Vercel Edge, unlimited on Cloudflare Workers)
• Memory limits (128MB typical)
• Subset of Web APIs — fetch, crypto, TextEncoder, URL, Request, Response are available

When to Use the Edge

• Personalisation and A/B testing at the CDN layer (before the page is served)
• Authentication and access control (JWT validation at the edge)
• Geolocation-based routing (serve different content by country)
• API rate limiting and bot protection
• Short, latency-sensitive computations that don't need a database round-trip