Workload shaping for QoS and power efficiency of storage systems

Abstract

The growing popularity of hosted storage services and shared storage infrastructure in data centers is driving the recent interest in resource management and QoS in storage systems. The bursty nature of storage workloads raises significant performance and provisioning challenges, leading to increased resource requirements, management costs, and energy consumption. We present a novel dynamic workload shaping framework to handle bursty server workloads, where the arrival stream is dynamically decomposed to isolate its bursty, and then rescheduled to exploit available slack. An optimal decomposition algorithm RTT and a recombination algorithm Miser make up the scheduling framework. We evaluate this framework using several real world storage workloads traces. The results show that workload shaping: (i) reduces the server capacity requirements and power consumption dramatically while affecting QoS guarantees minimally, (ii) provides better response time distributions over non-decomposed traditional scheduling methods, and (iii) decomposition can be used to provide more accurate capacity estimates for multiplexing several clients on a shared server.