Dynamic swarm segmentation

A method, implemented at a computer system that include one or more processors, for using a tree structure representing a dataset to identify a client peer for obtaining a portion of the dataset, the method comprising:

• identifying a dataset comprising a plurality of byte ranges;
• creating a tree structure that represents different subsets of the plurality of byte ranges of the dataset, the tree structure comprising a plurality of nodes that is each associated with a different subset of one or more ranges of the plurality of byte ranges, the plurality of nodes being hierarchically-arranged such that one or more byte ranges associated with each child node in the tree structure are a subset of one or more byte ranges associated with each parent node in the tree structure;
• identifying, for each client of a plurality of clients, one or more byte ranges of the plurality of byte ranges of the dataset that the client needs to obtain;
• associating each client of the plurality of clients with one of the plurality of nodes of the tree structure, including associating each client with a node of the tree structure that is, itself, associated with one or more byte ranges that match the one or more byte ranges that the client needs to obtain;
• and
• identifying, for at least a first client of the plurality of clients, at least one second client of the plurality of clients that is a peer to the first client for obtaining at least one byte range of the dataset, the identification being based on traversing the tree structure from a first node associated with the first client to a second node associated with the second client.

A computer program product comprising one or more hardware storage devices having stored thereon computer-executable instructions that are executable by one or more processors to use a tree structure representing a dataset to identify a client peer for obtaining a portion of the dataset, the computer-executable instructions including instructions that are executable to cause the one or more processors to perform at least the following:

• identify a dataset comprising a plurality of byte ranges;
• create a tree structure that represents different subsets of the plurality of byte ranges of the dataset, the tree structure comprising a plurality of nodes that is each associated with a different subset of one or more ranges of the plurality of byte ranges, the plurality of nodes being hierarchically-arranged such that one or more byte ranges associated with each child node in the tree structure are a subset of one or more byte ranges associated with each parent node in the tree structure;
• identify, for each client of a plurality of clients, one or more byte ranges of the plurality of byte ranges of the dataset that the client needs to obtain;
• associate each client of the plurality of clients with one of the plurality of nodes of the tree structure, including associating each client with a node of the tree structure that is, itself, associated with one or more byte ranges that match the one or more byte ranges that the client needs to obtain;
• and
• identify, for at least a first client of the plurality of clients, at least one second client of the plurality of clients that is a peer to the first client for obtaining at least one byte range of the dataset, the identification being based on traversing the tree structure from a first node associated with the first client to a second node associated with the second client.