A Novel Distributed Algorithm for High-Throughput and Scalable Gossiping

A family of gossiping algorithms depending on a combinatorial parameter is introduced, formalized, and discussed. Three members are analyzed. It is shown that, depending on the pattern of the parameter, gossiping can use from O(N2) to O(N) time, N being the number of communicating members. The last and best-performing algorithm, whose activity follows the execution pattern of pipelined hardware processors, is shown to exhibit high throughput and efficiency that are constantwith respect to N. This translates in unlimited scalability for the corresponding gossiping service provided by this algorithm