Learn the vocabulary of deliberately clearing stale cached content from a CDN's edge network.
0 / 5 completed
1 / 5
At standup, a dev mentions deliberately removing a specific cached asset from a CDN's edge servers immediately after publishing an updated version, so users don't keep receiving the stale copy. What is this action called?
Cache invalidation, or a purge, deliberately removes a specific cached asset from a CDN's edge servers immediately after it's updated, ensuring users receive the new version right away instead of continuing to receive a stale cached copy. Waiting passively for the cached asset's time-to-live to expire naturally means users could keep seeing outdated content for however long that TTL was configured. An explicit purge gives the team immediate, deliberate control over when stale content actually stops being served.
2 / 5
During a design review, the team wants to invalidate only the specific assets affected by a deployment, rather than purging the CDN's entire cache every time. Which capability supports this?
Targeted invalidation by path or tag purges only the specific assets actually affected by a given deployment, leaving unrelated cached content untouched and still served efficiently from the edge. Purging the entire cache indiscriminately on every deployment forces every single asset to be re-fetched from the origin server, creating unnecessary load and temporarily slower response times for content that never actually changed. This targeted approach keeps the CDN's caching benefit intact for everything that wasn't part of the actual update.
3 / 5
In a code review, a dev notices a cache invalidation request returns immediately but the actual purge across all edge locations completes asynchronously over the following seconds. What does this represent?
Eventually consistent cache purge propagation means an invalidation request is accepted immediately, but the actual removal of the cached asset from every distributed edge location takes a short additional time to fully complete. A cache purge that's instant and synchronous across every edge location everywhere isn't realistic given how geographically distributed a CDN's edge network typically is. Understanding this brief propagation delay matters for correctly reasoning about exactly when a purge is truly guaranteed to be complete everywhere.
4 / 5
An incident report shows a critical bug fix was deployed, but users in one region kept seeing the broken previous version for several minutes because the invalidation hadn't yet propagated to that region's edge servers. What practice would reduce this risk?
Accounting for the CDN's typical purge propagation delay when judging whether a critical fix has truly reached every user avoids prematurely assuming the issue is fully resolved everywhere. Assuming a purge completes instantly and synchronously everywhere overestimates what a globally distributed edge network can realistically guarantee in that short a time window. This awareness of propagation delay is especially important when verifying a critical, time-sensitive fix has actually taken full effect.
5 / 5
During a PR review, a teammate asks why the team invalidates only the specific changed assets after a deployment instead of just purging the CDN's entire cache every time, to keep things simple. What is the reasoning?
Purging the CDN's entire cache on every deployment forces every single asset, including ones that never changed, to be re-fetched from the origin server, creating unnecessary load and temporarily degraded performance. Targeted invalidation limits that re-fetch to only the assets actually affected by the specific change. The tradeoff is the small added complexity of tracking exactly which assets a given deployment actually touched, rather than just purging indiscriminately.