Power Modeling of Solid State Disk for Dynamic Power Management Policy Design in Embedded Systems

Abstract

Abstract.Power consumption now becomes the most critical performance limiting factor to solid state disk (SSD) in embedded systems. It is imperative to devise design methods and architectures for power efficient SSD designs. In our work, we present the first step towards low power SSD design, i.e., power estimation of SSD. We present a practical approach of SSD power estimation which tries to keep the advantage of real measurement, i.e., accuracy, while overcoming its limitations, i.e., long execution time and lack of repeatability (and high cost) by a trace-based simulation. Since it is based on real measurements, it takes into account the power consumption of SSD controller as well as that of Flash memories. We show the effectiveness of the presented method in designing a dynamic power management policy for SSD.

Design Tradeoffs of SSDs

“…Energy density of an SSD, that is, J/in 3 , is orders of magnitude larger than that of an HDD, and the peak energy consumption of an SSD is much higher than that of an HDD [Inoue et al 2011], which raises the heat dissipation issue. A number of works are dedicated to regulating the peak energy consumption via throttling the data transfer rate or limiting the number of flash memory chips programmed in parallel [Park et al 2009b;Lee et al 2013]. Compared to HDDs, SSDs will make CPU spend less time waiting for an I/O.…”

“…They propose that there should be an interface to inform the SSD of its Power Budget and that the firmware of an SSD should be designed to dynamically adjust the parallelism degree subject to its Power Budget. Several works have proposed dynamically throttling the transfer rate or parallelism degree of an SSD to regulate the temperature of the SSD [Park et al 2009b;Lee et al 2013]. To properly design the SSD internal parallelism, we need to incorporate the channel switch delay, way switch delay, and page write latency.…”

“…SSDs have been widely perceived as a means to deliver energy-efficient computer systems, replacing HDD-based storage systems [Poess and Nambiar 2010]. Recently, SSD vendors have adopted aggressive internal parallelism to boost the I/O performance of SSDs.…”

Power-reduction techniques for data-center storage systems

“…Because an SSD has no mechanical parts and is based on low-power NAND flash memory, it is more energy efficient than a regular disk drive [Park et al 2009]. An SSD does not require spin power nor seek power because it is composed of electronic components only.…”

“…Solid-state disks may also provide a standby mode in which electronic components are turned off to enable DPM [Park et al 2009]. The power-state transition costs are much lower for an SSD than an HDD because there's no disk that needs to spin down and spin up.…”