Robots and AI will run data centres
Artificial intelligence (AI) and blockchain are among new technologies that are driving a need for increased data centre capacity, according to a telco, announcing an expansion recently.
China Telecom said in a press release that these “rapidly maturing” technologies, such as machine learning and adaptive security, will propel investment in data centres. And that they are one reason for its data centre-business enlargement.
Interestingly, though, data centres themselves may end up using this new tech as heavily as the customers.
San Jose, California-based Litbit said in a recent blog post that it has developed the first AI-powered data centre operator.
AI tool helps prevent DC disasters
Dac, as the AI tool is called, promises to find loose electrical hook-ups and leaking water, among other potential data centre disasters. It uses machine learning.
Infrared vision is among Dac’s skills, says CEO Scott Noteboom, writing in the company’s blog. That “superhuman” insight, he claims, helps identify electrical arc flashes and alert managers to failing power supplies. Such things can be a precursor to server failure.
Litbit uses a human-to-machine learning interface that combines what existing human employees know and can tell the machines to look out for, along with real-time data. Vision, acoustics and touch combined with algorithms are used to detect anomalies.
“Clone your best employees,” the company’s marketing proclaims. In this case, the AI computer is taught about data centres by the incumbents. A manager oversees it, but Dac can process “thousands of data centre specific thoughts per second.”
Vibrations in server racks, for example, indicate hard drive issues—the drives exhibit acoustics that are not normal. Those anomaly sounds are compared to normal ones captured ultrasonically at varying load levels.
Environmental controls are ripe for AI management. HVAC, for example, can be adjusted based on weather.
Environmental controls are another feature ripe for AI management and are used by Dac—HVAC, for example, can be adjusted based on weather.
Noteboom qualifies the obvious human job-loss element of his product by saying his AI tool has been created to let data centre workers “focus on more interesting and new things.”
Additionally, however, Noteboom (speaking on the Infrastructure Masons panel at the DCD Webscale conference in San Francisco last month) explains that the Dac product would be great for data centre installs in remote locations, where highly skilled workers would have to be imported ordinarily. That is according to Data Centre Frontier, who wrote about the panel.
Enter the robot
Robots are another entrant in the data centre employment pool, says company Wave2Wave. The company’s product is a rack-mounted robot for making physical optical connections. The idea is to let data centres rapidly provision circuits.
Its tool, called Robotic Optical Switch for Data Centres (ROME) performs connections in a few seconds. The robot manages the cable connections, plucking components mechanically, almost like an old-school telecommunications switchboard where the panel-inserted electrical cords established voice connections.
It is faster than software-defined networking (SDN) orchestration software, the company’s website explains.
Advantages to both Litbit and Wave2Wave’s products are also found in security. Omitting humans from the data centre can be more secure.
“It’s a game changer in the physical connectivity space that’s overdue for innovation,” Wave2Wave says of its ROME connection-making robot.
IDG News Service