An algorithm for storing information redundantly on the nodes of a broadcast network is proposed. A voting technique is used to increase reliability. Since multiple votes are cast only when copies of a data item disagree, the algorithm has the property that communication overhead is minimal. In addition, nodes storing erroneous copies are automatically resynchronized. A Markov analysis is performed which relates parameters of the algorithm to the mean time to failure.
No takes yet. Share an insight, caveat, or question.
A. J. Bernstein (1985) studied this question.
Synapse has enriched 5 related papers. Consider them for comparative context: