According to an article in CampusTechnology online, Purdue University IT staff have developed methodologies in order to control the operational performance of machines based on the current data center heat load. Over the course of a sweltering summer, the Rosen Center for Advanced Computing at Purdue experienced several planned and unplanned outages. As a result, the IT staff had to develop a method to allow users access throughout the chiller downtime.
Power outages are actually infrequent at the data center, [Patrick] Finnegan said. But he added that this summer, “due to some planned cooling system maintenance, coupled with the unusually hot summer, we have had some brief cooling outages.”
In both instances, the cause was a temporary capacity reduction in the campus chilled water supply.”
Temperature sensors in the data center kick off scripts that throttle the relative power utilization of the machines. The article doesn’t go into technical specifics, but my geek radar says they’re likely tickling the CPU stepping via ACPI.
The program worked, and the datacenter didn’t overheat, so the process was a success. We actually were a bit surprised it worked so seamlessly,” said [Mike] Shuey. “It’s much better to have jobs run slowly for an hour than to throw away everyone’s work in progress and mobilize staff to try to fix things.”
For more info, check out their full article here.