Now Smith spends his days watching a dashboard to the Windows Servers running Big Data analytics on all that sensor data in real time. Anomalies set off an alarm that pinpoints the problem in minutes, enabling Microsoft to correct problems and fine-tune efficiencies that have cut its energy bills enough to pay off the cost of installing the software and input/output systems within 18 months after installation.
"All that sensor data is streaming into a SQL servers where the analytics provides a software overlay for all that operational data that we already had," said Smith. "Many of our buildings are now running 20 percent more efficient than before, and almost half the time any errors detected can be corrected from the console."
Most problems end up being related to incorrect set points -- which was the issue with that building that was running heating and air-conditioning at the same time. The turn-on set point for air-conditioning was set below the turn-off set point for the heating system. Of course, hardware malfunctions still have to be repaired, but the repairmen already know what the problems is and what tools or replacement parts they need to bring with them to make the repair, making even hardware fixes more efficient.
A wonderfully "green" use of real-time monitoring and distributed sensor networks. The implementation was truly ecologically friendly because Microsoft did not create waste in replacing their curren system; rather, they were able to make use of all the hardware and sensors they aleady had.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.