Improving Software Middleboxes and Datacenter Task Schedulers

Abstract

Shared systems have contributed to the popularity of many technologies. However, these systems often confront a common challenge: to ensure that resources are fairly divided without compromising utilization efficiency. In this master's thesis we look at this problem in two distinct systems---software middleboxes and datacenter task schedulers. We first present Sprayer, a system that uses packet spraying to load balance packets to cores in software middleboxes. Our design eliminates the imbalance problems of per-flow solutions and addresses the new challenges of handling shared flow states that come with packet spraying. Then, we present Stateful Dominant Resource Fairness (SDRF), a task scheduling policy for datacenters that looks at past allocations and enforces fairness in the long run. SDRF reduces users' waiting time on average and improves fairness by increasing the number of completed tasks for users with lower demands, with small impact on high-demand users.