Study Says Server Failures Don’t Rise With Heat

Servers aren’t as affected by the heat as you may have initially thought, according to a new study from the University of Toronto, which utilized data from thousands upon thousands of listed equipment failures from data centers run by

Los Alamos National Labs, Google, and even Canada’s SciNet. The study has provided some of the most expansive data the business has ever seen; the penalties behind high temperatures within the data center might not be as big as we once thought.

The study focuses on a select few problems regarding internal equipment malfunction (as opposed to server racks and mounting hardware) and observes the environmental conditions that the equipment was placed in. “Based on our study of data spanning more than a dozen data centers at three different organizations, and covering a broad range of reliability issues, we find that the effect of high data center temperatures on system reliability are smaller than often assumed,” the authors write. “For some of the reliability issues we study, namely DRAM failures and node outages, we do not find any evidence for a correlation with higher temperatures (within the range of temperatures in our datasets).”

For many errors that show a connection between temperature and failure rate, the relation is much weaker than expected. Internal temperatures below 50C showed that failures grew in a linear progression, as opposed to an exponential one, as many have suggested over the past years. The University believes that its results could allow for many more businesses and organizations to run at a higher temperature and operate with much more efficiency.
Thanks to findings such as these, many massive Internet companies such as Microsoft, Intel, and Google, have been aggressively turning up the temperatures, pushing the insides of many of their data centers above the 80 degree mark.