Distributed P2P quantile tracking with relative value error

TL;DR

This paper introduces DUDDSketch, a decentralized gossip-based algorithm for accurate quantile tracking in P2P networks, with proven correctness and convergence to sequential results.

Contribution

It presents a novel distributed quantile tracking algorithm that is fully decentralized and formally proven correct, suitable for unstructured P2P networks.

Findings

Algorithm converges to sequential results

Proven correctness of the distributed protocol

Effective in unstructured P2P networks

Abstract

In this paper we present \textsc{DUDDSketch}, a distributed version of the \textsc{UDDSketch} algorithm for accurate tracking of quantiles. The algorithm is a fully decentralized, gossip-based distributed protocol working in the context of unstructured P2P networks. We discuss the algorithm's design and formally prove its correctness. We also show, through extensive experimental results, that the algorithm converges to the results provided by the sequential algorithm, which is a fundamental and highly desirable property.