Method for streaming multimedia information over public networks

In a network having a content server which hosts streaming multimedia (SM) objects, each of said SM objects comprising a plurality of time-ordered packets for distribution over said network through a plurality of HSs to a plurality of clients, a method of reducing latency associated with distributing said SM objects from said content server and said plurality of helpers (HS) to said plurality of clients, said method comprising:

• receiving a first request for an SM object, including a requested starting position, received from one of said plurality of clients at one of said plurality of HSs;
• allocating a first ring buffer in a memory associated with said one of said plurality of HSs upon receiving said first request;
• retrieving said SM object comprising said plurality of time-ordered packets from at least one of said plurality of HSs including said one of said plurality of HSs and said content server;
• sequentially storing said plurality of time-ordered packets from said retrieved SM object in said first ring buffer by replacing lower time-ordered packets with higher time-ordered packets;
• servicing a second request for said SM object from said first ring buffer, if it is determined that said subsequent request includes a starting request position within a range between a largest timestamp associated with a packet stored in the buffer and a smallest time-stamp associated a packet stored in the buffer;
• and
• allocating a second ring buffer in the memory, if it is determined at the servicing step that said second request cannot be serviced from said first ring buffer.

A method of reducing latency in a network having a content server which hosts streaming media (SM) objects which comprise a plurality of time-ordered segments for distribution over said network through a plurality of helpers (HSs) to a plurality of clients, said method comprising:

• receiving a request for an SM object from one of said plurality of clients at one of said plurality of helper servers;
• allocating a buffer at one of said plurality of HSs to cache at least a portion of said requested SM object;
• downloading said portion of said requested SM object to said requesting client, while concurrently retrieving a remaining portion of said requested SM object from one of another HS and said content server;
• and
• adjusting a data transfer rate at said one of said plurality of HSs for transferring data from said one of said plurality of helper servers to said one of said plurality of clients.

A network of interconnected helper servers (HSs), each of said interconnected helpers comprising

• means for receiving and processing real-time streaming protocol (RTSP) requests from clients;
• means for forwarding client requests to a content server;
• means for streaming data to a plurality of clients using the real-time (RTP) protocol;
• means for managing available memory in the form of a buffer pool, each buffer in said buffer pool being associated with an SM object identified by a uniform resource locator (URL);
• means for mapping URLs identifying said objects received as a parameter along with an SM object request to local filenames;
• means for managing the disk space allocated for caching by implementing a cache replacement policy;
• means for recording data onto said cache and reading data from said cache;
• and
• scheduler means for managing data producer, data consumer, and garbage collector events, wherein said data producer events are events which source at least a portion of said SM object to said one of said plurality of helper servers to be stored in one of said buffers in said buffer pool, said data consumer events are RTSP requests from clients, and garbage collector events are events associated with removing a portion of said SM objects stored in said buffer pool to free the available memory.