Content Distribution & Caching Flashcards

A CDN edge node receives a 'Vary: Accept-Encoding' header from origin. What does this instruct the CDN to do?

Answer: Cache separate versions of the resource based on the Accept-Encoding request header value

The Vary header tells the CDN to store distinct cached copies for each unique value of the specified request header (e.g., gzip vs. br vs. identity).

What is 'cache stampede' (also called 'thundering herd') in CDN caching, and what is the primary mitigation?

Answer: When too many CDN nodes overload origin simultaneously; mitigated by request coalescing

Cache stampede occurs when a popular cached item expires and hundreds of simultaneous requests all miss and hit origin; request coalescing lets only one through.

Which HTTP/2 feature can CDNs leverage to push content to browsers before they request it, reducing round trips?

Answer: HTTP/2 Server Push

HTTP/2 Server Push allows CDN edge nodes to proactively send resources (CSS, JS) to the client alongside the initial HTML response.

A CDN is configured with 'stale-while-revalidate=60'. What happens during that 60-second window?

Answer: The CDN serves the stale cached response immediately while asynchronously fetching a fresh copy in the background

stale-while-revalidate lets the CDN return stale content instantly (zero latency penalty) while triggering a background refresh within the window.

Which content type is generally LEAST suitable for aggressive CDN caching with long TTLs?

Answer: Product listing pages with real-time inventory and pricing

Product pages with live inventory and pricing change frequently and vary per user, making long TTLs inappropriate without smart invalidation.

What is 'cache hierarchy collapse' and when is it desirable?

Answer: Removing all mid-tier caches to reduce latency when origin can handle the load

Collapsing the cache hierarchy (removing parent/shield tiers) reduces latency hops but increases origin traffic, suitable only for low-traffic or highly dynamic origins.

An e-commerce site uses a CDN and wants to cache product pages but show real-time cart counts. Which architecture achieves this correctly?

Answer: Cache the full product page HTML and use client-side JavaScript to fetch cart data from a separate non-cached API endpoint

Separating cacheable page content from dynamic per-user data (fetched via API after page load) is the standard pattern for caching personalized pages.