The technology is suitable for any kind of distributed networks having different data rates on different connections and a computer system where the data to distribute is created. The system comprises one or more supersets of data to be distributed to client computers by extracting subsets of data from the superset(s) of data. Thereafter, the subsets of data are distributed to selected groups of clients based on information about the network architecture. This provides an efficient way of distributing data without having to duplicate distributed data. In example implementations, subsets may be distributed to clients based on a counter performance or a data rate of a client's connection to the computer system.