DEMon: Decentralized Monitoring for Highly Volatile Edge Environments

Abstract

Monitoring systems play an essential role in efficiently managing resources and application workloads by collecting, storing, and providing requisite information about the state of the resources. However, traditional monitoring systems that collect and store the data in centralized remote servers are infeasible for Edge environments. The centralized architecture increases the communication latency for information storage and retrieval and creates a failure bottleneck. In addition, the Edge resources are arbitrarily (de)provisioned, which creates further challenges for providing quick and trustworthy data. Thus, it is crucial to design and build a monitoring system that is fast, reliable, and trustworthy for such volatile Edge computing systems. Therefore, we propose a Decentralised Edge Monitoring (DEMon) framework, a decentralized, self-adaptive monitoring for highly volatile Edge environments. DEMon, at the core, leverages the stochastic Gossip communication protocol and develops techniques for efficient information dissemination, communication, and retrieval, avoiding a single point of failure and ensuring fast and trustworthy data access. We implement it as a lightweight and portable container-based monitoring system and evaluate it through empirical experiments. The results show that DEMon efficiently decimates and retrieves the monitoring information, addressing the abovementioned challenges.