Open problems in queueing theory inspired by datacenter computing

Datacenter operations today provide a plethora of new queueing and scheduling problems. The notion of a "job" has become more general and multi-dimensional. The ways in which jobs and servers can interact have grown in complexity, involving parallelism, speedup functions, precedence constraints, and task graphs. The workloads are vastly more variable and more heavy-tailed. Even the performance metrics of interest are broader than in the past, with multi-dimensional service-level objectives in terms of tail probabilities. The purpose of this article is to expose queueing theorists to new models, while providing suggestions for many specific open problems of interest, as well as some insights into their potential solution.