Share on Facebook Tweet on Twitter Share on LinkedIn Share by email
The SCADS Director: Scaling a Distributed Storage System Under Stringent Performance Requirements

Beth Trushkowsky, Peter Bodík, Armando Fox, Michael J. Franklin, Michael I. Jordan, and David A. Patterson

Abstract

Elasticity of cloud computing environments provides an economic incentive for automatic resource allocation of stateful systems running in the cloud. However, these systems have to meet strict performance Service-Level Objectives (SLOs) expressed using upper percentiles of request latency, such as the 99th. Such latency measurements are very noisy, which complicates the design of the dynamic resource allocation. We design and evaluate the SCADS Director, a control framework that reconfigures the storage system on-the-fly in response to workload changes using a performance model of the system. We demonstrate that such a framework can respond to both unexpected data hotspots and diurnal workload patterns without violating strict performance SLOs.

Details

Publication typeInproceedings
Published inFAST '11: Conference on File and Storage Technologies
PublisherUSENIX
> Publications > The SCADS Director: Scaling a Distributed Storage System Under Stringent Performance Requirements