Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, are described for establishing a grouping rule and a mapping protocol at a cache server system for caching fragments of a media stream that includes a timed-sequence of video frames. The fragments can be generated and identified at media source systems in accordance with a fragmenting format and an indexing protocol, respectively, which are common across the media source systems. Further, the established mapping protocol can be different from the indexing protocol. The methods include retrieving at least a portion of the fragments of the media stream by the cache server system from the media source systems. The methods also include allocating the retrieved fragments to local fragment groups at the cache server system based on the established grouping rule. The methods further include mapping the allocated fragments to the local fragment groups based on the established mapping protocol.