The Splendors and Miseries of Rounds

Abstract

Take a stroll with me through the distributed computing literature: we could scour the proceedings of PODC, DISC, SIROCCO, OPODIS, and more; we could explore the issues of Distributed Com- puting; or we could skim through the multitude of preprints on arXiv. Yet, I can already predict that with good probability, the paper we choose to read will contain the word \round" or łayer".1 Is it a lower-bound paper? Then round complexity is probably one measure of performance considered. Is it a combinatorial topology paper? Almost all techniques developed from application of combinatorial topology assume layers of communication, such that an execution can be reduced as successive complexes; here again rounds are essential. Is it a paper on fault-tolerance? Then some broadcasting strategy to ensure decent replication probably uses rounds.