The Peer Content Caching and Retrieval: Retrieval Protocol reduces bandwidth consumption on branch-office wide-area-network (WAN) links by having clients retrieve content from distributed caches when available instead of the content servers, which are often located remotely from branch offices over the WAN links. It is based on a peer-to-peer discovery and distribution model, where the peers themselves act as caches from which they serve other requesting peers. The framework also supports the mode of using pre-provisioned hosted caches in place of peer-based caching. The main benefit of the framework is to reduce operation costs by reducing WAN link utilization, while providing faster downloads from the local area networks (LANs) in the branch offices.
The Retrieval Framework defines four protocol message exchanges: for querying the protocol version of the server, for querying the server for the availability of certain content (two message exchanges), and for retrieving content from a server. The framework incorporates both the Retrieval Protocol and the Discovery Protocol [MS-PCCRD] together to enable a client to discover and retrieve content from multiple peers that have the content instead of the original content server.
Sections 1.8, 2, and 3 of this specification are normative and can contain the terms MAY, SHOULD, MUST, MUST NOT, and SHOULD NOT as defined in [RFC2119]. Sections 1.5 and 1.9 are also normative but do not contain those terms. All other sections and examples in this specification are informative.