Fair Scheduling for Deadline Driven, Resource-Constrained Multi-Analytics Workloads

In this paper, we analyze and empirically evaluate Justice, a new approach to fair-share, deadline-aware job scheduling for resource-constrained cloud deployments managed by big data resource negotiators. Justice provides admission control, which leverages historical traces and job deadlines to guide and adapt resource allocation decisions to changing workload conditions. We evaluate Justice using different deadline types and production analytics workloads. We find that it outperforms extant allocators in terms of fair allocation, deadline satisfaction, and useful work.