Understanding the Operational Carbon Footprint of Storage Reliability and Management

Varsha Rao (University of Chicago); Andrew A. Chien (University of Chicago and Argonne National Laboratory)

Abstract

With daily data generation of Zettabytes and exponential growth, our study finds that power needed for data reliability and management is an increasing fraction of storage system power. Thus storage management is an important contributor to data center (DC) energy use and carbon footprint.

We study the University of Chicago’s high energy physics storage system (UChicago HEP) background tasks for reliability and management. We build a model for their activity and power costs, and next explore their opportunities for temporal shifting to reduce operational carbon footprint.

We apply our model to varied DC scale: a large HEP facility, large cloud, and global set of large cloud DCs, we find that these storage tasks consume 2.1-4.9% of DC power and 12.8-27.2% of DC storage annually. Translating power use to carbon emissions depends on location and time, so carbon foot- print varies. But we show that storage management power can account for as much as the annual electricity consumption of 755,000 US homes. Studying four power grids with varied carbon intensity, we show that aligning back- ground tasks with low-carbon periods can achieve emission reductions of up to 82.8%. Greater reductions (up to 96.9%) are possible, as grids decarbonize further.