1 / 1

energy

Temperature Aware Storage. George Goldberg, Ronen I. Kat and Dmitry Sotnikov. IBM Research - Haifa. Executive Summary.

liesel
Télécharger la présentation

energy

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Temperature Aware Storage George Goldberg, Ronen I. Kat and Dmitry Sotnikov IBM Research - Haifa Executive Summary Heat removal is a dominant activity in today's data centers. Despite the existing debate about the exact range of operational temperatures for informational technology equipment, heat needs to be removed. This activity can incur high energy consumption. Rotational Speed vs. Energy Data Center Cooling Costs Enterprise storage systems include dozens of fans. To improve the fan’s cooling efficiency vendors use fans with variable rotational speeds. Looking at rotational energy separately around an object's axis of rotation, one gets the following dependence on the object's moment of inertia: This means that to double the rotational speed, we need to multiply the energy by a factor of four • Cooling a server consumes 1 to 2 watts for every watt that powers it. • On average, 50% of total electricity usage goes towards cooling. Temperature as a function of time and workload Temperature as a Function of Time and Workload energy The temperature is afunction of two variables:time and workload More intensive workloadcauses faster heating andhigher heat removal rate. Increased disk temperature results in an increase in fans’ RPM, which consume additional power. The disk drives we used for testing, are designed to work at a specific temperature range. Violating this operational range may cause direct or indirect damage to the disk or data. Some disk manufacturers include a device level mechanism to prevent overheating by reducing head movement and speed. Hence, reducing the heat output incurs costs related to reduced performance. However, such devicelevel mechanisms canimpact overall systemperformance. Instead, we suggest tackling heat removal at the system andworkload levels. Temperature Aware Storage Management energy Devices that are equipped with safeguards against overheating, will not need additional cooling activities. As a system level placement policy, these safeguards will make sure that scenarios such as disk overheating due to excessive use, will not take place. Reducing Power Consumption by Data Migration energy Storage workloads usually have patterns and locality. Proper data placement can reduce temperature peaks. This results in a reduction in fan usage, and therefore, reduces system energy consumption. Hot Data Migration Reduction in Power Consumption Contacts: George Goldberg, Ronen Kat, Dmitry Sotnikov<georgeg, ronenkat, dmitrys@il.ibm.com>

More Related