ASAP: Adaptive and Scalable Microservice Provisioning for Edge-IoT Networks

Abstract

The edge computing paradigm aims at provisioning compute and storage resources from Internet-of-Things (IoT)-enabled devices at the network edge, while disseminating the end-to-end latency and increasing mobile computational capacity. For scalable provisioning of microservices, recently proposed methods frequently need a priori information on the microservice type, computational capacity, and other parameters. In the presence of such restrictions, the present methods need to face lower quality of experience for mission-critical and resource-hungry applications. In this paper, we study the problem of microservice provisioning for mobile edge computing, and we propose an adaptive and scalable solution. The core of our optimal microservice differentiation scheme is ASAP, a microservice-level abstraction for the orchestration of network resources. ASAP provides adaptive and scalable microservice provisioning by minimizing microservice delay, while maximizing the network throughput. With a prototype tested over a local testbed and trace-driven simulations, we show how ASAP increases throughput compared to other solutions based on the executions of microservices at the edge.