Robust aggregation protocols for large-scale overlay networks

摘要: Aggregation refers to a set of functions that provide global information about distributed system. These junctions operate on numeric values over the system and can be used count network size, determine extremal compute averages, products or sums. allows important basic functionality achieved in fully peer-to-peer networks. For example, monitoring application, some aggregate reaching specific value may trigger execution certain operations; storage systems need know total free space available; load-balancing protocols benefit from knowing target average load so as minimize transfered load. Building simple but efficient idea antientropy aggregation (a scheme based epidemic communication model), this paper we introduce practically applicable robust adaptive for proactive aggregation, including calculation average, product values. We show how averaging protocol applied further aggregates like sum, variance size. present theoretical empirical evidence supporting robustness under different scenarios.