Robust Aggregation Protocols for Large-Scale Overlay Networks

  • Authors:
  • Alberto Montresor;Mark Jelasity;Ozalp Babaoglu

  • Affiliations:
  • University of Bologna, Italy;University of Bologna, Italy;University of Bologna, Italy

  • Venue:
  • DSN '04 Proceedings of the 2004 International Conference on Dependable Systems and Networks
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

Aggregation refers to a set of functions that provideglobal information about a distributed system. These functionsoperate on numeric values distributed over the systemand can be used to count network size, determine extremalvalues and compute averages, products or sums. Aggregationallows important basic functionality to be achieved infully distributed and peer-to-peer networks. For example, ina monitoring application, some aggregate reaching a specificvalue may trigger the execution of certain operations;distributed storage systems may need to know the total freespace available; load-balancing protocols may benefit fromknowing the target average load so as to minimize the transferedload. Building on the simple but efficient idea of anti-entropyaggregation (a scheme based on the anti-entropyepidemic communication model), in this paper we introducepractically applicable robust and adaptive protocolsfor proactive aggregation, including the calculation of average,product and extremal values. We show how the averagingprotocol can be applied to compute further aggregateslike sum, variance and the network size. We present theoreticaland empirical evidence supporting the robustness of theaveraging protocol under different scenarios.