Massive arrays of idle disks for storage archives
Proceedings of the 2002 ACM/IEEE conference on Supercomputing
DRPM: dynamic speed control for power management in server class disks
Proceedings of the 30th annual international symposium on Computer architecture
Energy conservation techniques for disk array-based servers
Proceedings of the 18th annual international conference on Supercomputing
Power-Aware Storage Cache Management
IEEE Transactions on Computers
Hibernator: helping disk arrays sleep through the winter
Proceedings of the twentieth ACM symposium on Operating systems principles
Disk drive level workload characterization
ATEC '06 Proceedings of the annual conference on USENIX '06 Annual Technical Conference
ACM Transactions on Storage (TOS)
Write off-loading: practical power management for enterprise storage
FAST'08 Proceedings of the 6th USENIX Conference on File and Storage Technologies
Restrained utilization of idleness for transparent scheduling of background tasks
Proceedings of the eleventh international joint conference on Measurement and modeling of computer systems
Optimality analysis of energy-performance trade-off for server farm management
Performance Evaluation
iPOEM: a GPS tool for integrated management in virtualized data centers
Proceedings of the 8th ACM international conference on Autonomic computing
Saving disk energy in video servers by combining caching and prefetching
ACM Transactions on Multimedia Computing, Communications, and Applications (TOMCCAP) - Special issue of best papers of ACM MMSys 2013 and ACM NOSSDAV 2013
Hi-index | 0.01 |
Low utilization immediately suggests that placing the system into a low power mode during idle times may considerably decrease power consumption. As future workload remains largely unknown, "when" to initiate a power saving mode and for "how long" to stay in this mode remains a challenging open problem, given that performance degradation of future jobs should not be compromised. We present a model and an algorithm that manages to successfully explore feasible regions of power and performance, and expose the system limitations according to both measures. Extensive analysis on a set of enterprise storage traces shows the algorithm's robustness for successfully identifying "when" and for "how long" one should activate a power saving mode given a set of power/performance targets that are provided by the user.