Data centres are good places to chill-out… literally
ARTIFICIAL INTELLIGENCE KNOWS HOW TO CHILL… servers; or at least Google’s AI tech can, as it’s being used to handle data centre cooling systems.
AI from Google’s DeepMind division is being given the task of controlling the cooling in Google’s data centres with an aim of trying to reduce energy consumption while stopping servers from frying while they support the likes of Google Search, YouTube and Gmail.
“In 2016, we jointly developed an AI-powered recommendation system to improve the energy efficiency of Google’s already highly-optimised data centres. Our thinking was simple: even minor improvements would provide significant energy savings and reduce CO2 emissions to help combat climate change,” said DeepMind’s boffins.
“Now we’re taking this system to the next level: instead of human-implemented recommendations, our AI system is directly controlling data centre cooling, while remaining under the expert supervision of our data centre operators. This first-of-its-kind cloud-based control system is now safely delivering energy savings in multiple Google data centres.”
Said supervision will be carried out by fleshy human meatbags, meaning the AI isn’t doing people out of jobs.
The AI aims to boost efficiency by taking a snapshot of the cooling system of a data centre from thousands of sensors and pushing it through a deep learning neural network. That network works to predict how different combinations of potential actions will affect the future energy consumption of the cooling system.
“The AI system then identifies which actions will minimise the energy consumption while satisfying a robust set of safety constraints. Those actions are sent back to the data centre, where the actions are verified by the local control system and then implemented,” explained the DeepMind folks.
Apparently, the AI use so far has resulted in the ability for Google’s data centre operators to implement more granular actions to improve efficiency while making fewer mistakes doing so.
The AI can also predict levels of uncertainty in the potential decisions it could make, so if an action doesn’t have a good degree of confidence that it’ll yield good results, it can be scrapped.
And two-layer authentication, whereby AI actions are vetted by an internal list of safety constraints created by data centre operators, prevent the AI system from making a decision that could bork the data centres it’s looking after.
Managing cooling systems might not be the most exciting use of AI systems – for that look at driverless cars or the work to make AI help with healthcare. But given how data centres gobble up energy, the more efficient they can be the smaller their carbon footprint, which should prevent baby cows from coughing and help other eco-friendly stuff. µ
Source : Inquirer