Share on Facebook Tweet on Twitter Share on LinkedIn Share by email
Jockey: Guaranteed Job Latency in Data Parallel Clusters

Andrew Ferguson, Peter Bodik, Srikanth Kandula, Eric Boutin, and Rodrigo Fonseca

Abstract

Data processing frameworks such as MapReduce and Dryad are used today in business environments where customers expect guaranteed performance. To date, however, these systems are not capable of providing guarantees on job latency because scheduling policies are based on fair-sharing, and operators seek high cluster use through statistical multiplexing and over-subscription. With Jockey, we provide latency SLAs for data parallel jobs written in SCOPE. Jockey precomputes statistics using a simulator that captures the job's complex internal dependencies, accurately and efficiently predicting the remaining run time at different resource allocations and in different stages of the job. Our control policy monitors a job's performance, and dynamically adjusts resource allocation in the shared cluster in order to maximize the job's economic utility while minimizing its impact on the rest of the cluster. In our experiments in Microsoft's production Cosmos clusters, Jockey meets the specified job latency SLAs and responds to changes in cluster conditions.

Details

Publication typeProceedings
Published inEuroSys
> Publications > Jockey: Guaranteed Job Latency in Data Parallel Clusters