Energy Dumpster Diving
Published in Second Workshop on Power Aware Computing (HotPower), November 2009.
Power data alone cannot identify sources of energy inefficiency. However, correlating power data with utilization statistics can reveal where power is used well and where it is wasted. We describe a sensing infrastructure, PowerNet, that monitors power and utilization in a building environment. The deployment includes both wired and wireless sensors and covers offices, networking closets, and server racks. We present PowerNet's architecture, then generate initial insights from each monitored environment. Analyzing PowerNet data traces identifies contexts where electricity consumption can be reduced without cost, and others which call for rethinking system designs altogether.
Talk (26MB), Paper (816KB)
BibTeX entry
@inproceedings{hotpower09kaz, author = "Maria Kazandjieva and Brandon Heller and Philip Levis and Christos Kozyrakis", title = "{Energy Dumpster Diving}", booktitle = "{Second Workshop on Power Aware Computing (HotPower)}", year = {2009}, month = {November} }