Robust and efficient membership management in large-scale dynamic networks

Abstract

Epidemic protocols are a bio-inspired communication and computation paradigm for large-scale network system based on randomised communication. These protocols rely on a membership service to build decentralised and random overlay topologies. In large-scale, dynamic network environments, node churn and failures may have a detrimental effect on the structure of the overlay topologies with negative impact on the efficiency and the accuracy of applications. Most importantly, there exists the risk of a permanent loss of global connectivity that would prevent the correct convergence of applications. This work investigates to what extent a dynamic network environment may negatively affect the performance of Epidemic membership protocols. A novel Enhanced Expander Membership Protocol (EMP+) based on the expansion properties of graphs is presented. The proposed protocol is evaluated against other membership protocols and the comparative analysis shows that EMP+ can support faster application convergence and is the first membership protocol to provide robustness against global network connectivity problems.

Accuracy of Distributed Systems Towards Industry 4.0: Smart Grids and Urban Drainage Systems Case Studies

“…Epidemic membership protocols offer mechanisms intended to maintain the connectivity of the systems in terms of overlay topology [32]. Each sensor holds the local cache containing the connections to 20 random sensors.…”

An Adaptive Restart Mechanism for Continuous Epidemic Systems

“…Each node is assigned a unique identifier and communicates with random peers that are selected uniformly from the system (a.k.a uniform gossip [6]). The Node Cache Protocol (NCP) [14] is a simple peer-sampling service adopted for the purpose of this work, although any other membership protocol with better properties can be used (e.g., [10]).…”

“…Theoretical and practical analyses have shown stochastic guarantees on the convergence to the desired target of stable epidemic systems [6,7]. Achieving and detecting convergence is critical for many services: for example, it is a prerequisite for actions such as termination and restart [1,10]. Moreover, real-world systems are generally asynchronous and highly dynamic.…”

“…Also, nodes usually have access to different clocks, encounter variable communication delays and may lose messages or message order. In dynamic systems, nodes can join and leave arbitrarily (a.k.a 'Node Churn') [10], and may suddenly fail or become unreachable [1].…”

“…Node churn and unreliable networks have a detrimental effect on the efficiency and the robustness of epidemic systems [10,11]. In extreme case, the inherent properties of the epidemic systems cannot be guaranteed: tasks may not converge or may converge to incorrect results.…”