HTTP Delivery and Cache Ownership

Match response reuse, variants, and resource priority to public and private content.

The browser, a shared edge cache, and an application server may all keep copies of a response. Those copies have different audiences and lifetimes. The same URL can also have language or encoding variants; a cache must know which request fields changed the representation. Even correct caching can feel slow when a critical image is discovered late or a low-value script competes for bandwidth. These lessons connect response ownership, stale reuse, content negotiation, and resource priority to one review application without assuming that faster transfer alone makes the page correct.

Topics in this track

Shared Cache Keys and Private Response Boundaries — Choose cache ownership and variant keys before putting response bytes at an edge.
Stale Revalidation and Versioned Purges — Use stale responses only where delayed freshness is acceptable and invalidation is observable.
Content Negotiation and Compression Contracts — Serve the right representation and keep caches aware of language and content encoding.
Critical Resource Discovery and Priority — Make the first important image or text resource discoverable early without flooding the network.