Title:Redundancy Management for Fast Service (Rates) in Edge Computing Systems

Abstract:Edge computing operates between the cloud and end-users and strives to provide fast computing services for multiple users. Because of their proximity to users, edge services have a low communication delay and can provide low latency with sufficient computing and storage resources. However, edge computing and storage resources are limited. Thus, directing more resources to some computing jobs will block (and pass to the cloud) the execution of others. We evaluate the edge system performance using two metrics: job computing time and job blocking probability. Edge nodes often operate in highly unpredictable environments and handle jobs needing fast or no service. Thus, jobs not getting into service upon arrival get blocked and passed to the cloud. In unpredictable environments, replicating a job to multiple servers when resources allow shortens its computing time. However, such replication makes the resources unavailable to other users, and their execution is blocked. We show that the job computing time decreases with increasing replication factor, but the job blocking probability does not. Therefore, there is a tradeoff. This paper uses the average system time and service rates as performance metrics to evaluate the tradeoff. We conclude that the optimal number replication factor that minimizes the average system time changes with the distribution parameters and the arrival rate.