Predictive Prefetching reduces the page load time by ensuring, with high probability, that an object embedded in the page is available on the edge server, before the user's request.
Unlike regular prefetching, Predictive Prefetching:
- uses historical data to predict which objects will be required to render a page when a request for that page is received,
- occurs before the edge server receives the requested page from the origin, thereby saving a round trip to the origin,
Predictive Prefetching is completed on the edge server that receives the parent request. The edge server:
- Receives a request from an end user (the parent request) that is enabled for predictive prefetching.
- Analyzes the parent request to determine which objects are related to the parent request.
- Records statistics about the parent request.
- Rates the related dependent objects, based on the URL of the parent request and other request parameters.
- Applies the prefetch level to determine whether to prefetch one or more dependent objects.
- Fetches dependent objects from the source while the parent request is sent to the origin.
- Sends the fetched dependent objects to the client when requested.
|What it does
|Enables or disables the behavior.
|Select which prefetch level to apply.
Consider that a higher level can yield higher page performance but may also lead to higher midgress costs and potentially higher load on your origin.
Updated over 2 years ago