Request Collapsing

As soon as end-user requests hit our edge servers, the cache is checked for the requested content. In cases where the content is not cached (MISS) on any layer of our multi-layer cache infrastructure the requests will be forwarded from our origin shield layer to your origin.

In normal circumstances, each cache MISS request would be forwarded to the origin, prompting the content to be downloaded to the CDN77 cache servers per request until the download had finished, allowing requests made subsequent to the cache fill to be served directly from cache.

This is where request collapsing comes in. Request collapsing analyzes the requests that make it to our origin shield layer and temporarily "holds" matching requests, allowing the fewest number of matching requests to be forwarded to the origin as possible. As long as the initial download from the origin server finishes within the duration of the "cache lock", the held requests are registered as cache HIT and delivered directly from the cache. In case the request takes longer than the set duration, those requests will go to the origin server, registering as cache MISS.

This further utilises CDN77’s cache to ensure the fewest amount of requests are forwarded to your origin, reducing load and egress.