Периферийные вычисления, облако и CDN

Cloud, CDN, and edge all chase fast, reliable apps — they just place your compute and data in different spots. Knowing which is which is the whole game.

In the classic cloud model your code runs in a few central regions. Plenty of power, but distant users eat the latency.

A CDN caches static assets at hundreds of locations near users. Great for speed — but traditionally it only serves static content, not your logic.

Edge computing runs your actual code at those near-user locations. You get CDN-level proximity plus dynamic compute — the best of both.

For a Tokyo user hitting a US region, cloud is ~150ms round trip while edge is ~10-20ms to the nearest node. For latency-sensitive logic, edge wins.

Edge runtimes like Cloudflare Workers use lightweight isolates, giving near-zero cold starts — far quicker than spinning up serverless containers.

Edge trades power for proximity: limited CPU time per request, smaller memory budgets, and restricted APIs — no full filesystem. Plan around it.

Real apps blend all three: CDN for assets, edge for auth, routing, and personalization, and cloud for heavy databases and batch jobs.

Reach for edge when you need A/B testing and personalization, auth and request routing, or API aggregation close to your users.

Keep heavy compute, large stateful databases, GPU workloads, and long jobs in the cloud, where resources are abundant and limits are generous.

Think of it as a spectrum: static-only is CDN, dynamic-and-near-user is edge, heavy-and-central is cloud. Most production systems blend all three.

Recap: you compared cloud, CDN, and edge — edge's latency and cold-start wins, its resource limits, and how hybrid architectures use each where it fits.