A Provably-Efficient Online Algorithm for Re-Utilizing Unused VM Resources for Edge Providers

Qihang Sun, Shaolei Ren, Chuan Wu

Abstract: In recent years, as a result of the rapidly growing volume of generated data, computation has been increasingly migrating from megascale data centers to Internet edges (a.k.a edge computing), for avoiding high latencies and overwhelmed bandwidths. Unlike centralized clouds, edge computing processes workloads generated by users nearby. Thus, due to the lack of statistical multiplexing from a large group of users, the resource demand at an edge data center exhibits more fluctuations, resulting in time-varying unused computation resources. In this paper, we propose our UNusEd spAred VM Re-uTilizing mecHanism, UNEARTH, to utilize different types of unused resources, such as storage, CPU, GPU, and so on, offered by an edge computing provider. Notably, the exact amount of unused VM resources is unknown before selling them. We evaluate the performance of our algorithms under realistic settings, showing that our proposed VM bundle allocation algorithm can achieve (1+ Ω/Ω-1 ε (e)M <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1/Ω-1</sup> -1)-approximation in the worst case compared with optimums; and overall, our algorithms outperform the existing and heuristic algorithms.