Distributed Resource Selection for Self-Organising Cloud-Edge Systems

TL;DR

The paper tackles scalable resource selection in the Cloud-Edge continuum by replacing a central controller with a distributed CBBA-based mechanism among Resource Agents. It juxtaposes a centralised baseline (exhaustive or First-Fit) with the CBBA approach, detailing bidding/consensus, convergence guarantees, and practical adaptations for cloud-edge constraints. Empirical results show CBBA achieves near-optimal allocations with substantially lower computation time and strong scalability, outperforming centralised heuristics in large-scale scenarios while maintaining QoS and cost parity. This work demonstrates a viable path toward fully decentralised, self-organising orchestration capable of handling dynamic, heterogeneous infrastructures in real time.

Abstract

This paper presents a distributed resource selection mechanism for diverse cloud-edge environments, enabling dynamic and context-aware allocation of resources to meet the demands of complex distributed applications. By distributing the decision-making process, our approach ensures efficiency, scalability, and resilience in highly dynamic cloud-edge environments where centralised coordination becomes a bottleneck. The proposed mechanism aims to function as a core component of a broader, distributed, and self-organising orchestration system that facilitates the intelligent placement and adaptation of applications in real-time. This work leverages a consensus-based mechanism utilising local knowledge and inter-agent collaboration to achieve efficient results without relying on a central controller, thus paving the way for distributed orchestration. Our results indicate that computation time is the key factor influencing allocation decisions. Our approach consistently delivers rapid allocations without compromising optimality or incurring additional cost, achieving timely results at scale where exhaustive search is infeasible and centralised heuristics run up to 30 times slower.

Figures (7)

• Figure 1: Allocation time (in log-scale) by application with varying number of capacities (ranging from 10 to 15). Black dotted lines represent the five repetitions.
• Figure 2: Number of failed allocations for $50$ applications under varying levels of capacities.
• Figure 3: Allocation cost for $50$ applications.
• Figure 4: Difference of allocation costs for individual applications with $12$ available capacities.
• Figure 5: QoS values across $50$ applications for all capacities ($10$ to $15$). Lower value means better QoS.