Edge computing has become increasingly popular across many domains and enterprises. However, given the locality constraint of edges (i.e., only close-by edges are useful), multiplexing diverse workloads becomes challenging. This results in poor resource utilization in edge resources that are provisioned for peak demand. A simple way to allow multiplexing is through micro-data centers, that bring computation close to the users while supporting diverse workloads throughout the data, along with edges. In this paper, we argue for a hybrid approach of dedicated edge resources within an enterprise and on demand resources in micro-data centers that are shared across entities. We show that this hybrid approach is an effective and cost-efficient way for scaling workloads and removes the need for over-provisioning dedicated resources per enterprise. Moreover, compared to a scaling approach that uses only the edges across enterprises, micro-data centers also form a trusted third party that can maintain important quality of service guarantees such as data privacy, security, and availability.
@article{arxiv.1806.09265,
title = {Ben\'e: On Demand Cost-Effective Scaling at the Edge},
author = {Faria Kalim and Shadi A. Noghabi},
journal= {arXiv preprint arXiv:1806.09265},
year = {2018}
}