Power-Efficient Server-Class Performance from Arrays of Laptop Disks

Data centers capable of providing Internet, application, database and network services are an increasingly important component of the world’s computing infrastructure. In 1995 there were 20,000 servers in the world. As of June 2001, that number had reached six million [5]. Most existing research on data center design has aimed to improve performance, reliability and availability. Recently, however, researchers have begun to recognize the importance of energy efficiency [1, 2]. Increased data center power consumption translates directly into higher total cost of ownership, attributable to operating power, cooling, and decreased reliability. The disk array of a server-class system can account for a significant portion of the server’s total power budget. A recent white paper suggests that disk drives in a data center can account for 27 % of total electric consumption [5]. In some configurations the fraction can be significantly higher. A Dell PowerEdge 6650 [3], for example, comes equipped with 4 Intel Xeon 2.0 GHz processors and 292 15 KRPM hard drives. The processors are rated at 58W each, while an operational SEAGATE ST318453 15KRPM 18GB server-class disk drive consumes 15W. In such a configuration the hard disks consume 15 times more power than the processors. Similar observations for mobile (e.g. laptop) systems have led to the development of power management policies that spin down the hard disk when it is idle, but these policies do not transfer well to server-class disks. Serverclass disks are characterized by higher power mode transitioning costs, both in terms of power and latency, while server workloads are significantly more data intensive, leading to very short idle periods that cannot be exploited efficiently by non-operational low power modes. Adaptive Throughput To improve the energy efficiency of server-class storage systems Gurumurthi et al. have suggested the use of DRPM [4], an approach that dynamically modulates disk speed depending on current workload, decreasing th