Automated Storage Management with QoS Guarantee in Large-scale Virtualized Storage Systems

Storage virtualization in modern storage systems allows variability in the number of "physical" disks supporting a single "virtual" disk. In practice, IO workloads vary with time. Presuming the evolution of reasonable predictive models with the power of accurately predicting IO workload, it may be argued that it is straightforward to compute the number of disks needed at any time to satisfy QoS constraints. However, because disks contain data, the data also needs to be reallocated amongst disks. Not only does data migration need to be scheduled ahead but it must also be scheduled in such a way that QoS violations do not occur because of the extra migration IOs. In this paper, we present a novel analytic framework, PULSTORE, for autonomically managing the storage to balance both cost and performance. Given the workload characteristics of an application and storage QoS requirements, our PULSTORE framework yields an optimal schedule for adding and removing disks to su...

• Type: Journal
• Year: 2006
• Where: DEBU
• Authors: Lin Qiao, Balakrishna R. Iyer, Divyakant Agrawal, Amr El Abbadi
• Keywords: DEBU 2006, IO Workload, Modern Storage Systems, Reasonable Predictive Models