September 14, 2018
By Paul Barker
No company can survive if it stagnates and nowhere is this more true than in the data centre operation space where competition is fierce.
AJ Byers, former president of Rogers Data Centers and now president and CEO of Montreal-based Root Data Centers, says the key to success lies first in the ability to quickly build out capacity and secondly the adoption of artificial intelligence or AI in operations.
On the AI front, the company, which in June announced the expansion of its MTL-R2 facility with the addition of two new data halls, has been aggressively implementing new and interesting ways of avoiding dreaded downtime.
Last December, the company says it became the world’s first wholesale data centre in the world to use AI and machine learning through the signing of a technology agreement with Litbit, a San Jose, Calif. company that has developed a platform targeted at personnel that manage spaces and equipment.
“Business is everywhere and always in the data centre more than anywhere else, but people can’t be,” said Litbit CEO Scott Noteboom at the time. “That is the promise of AI, to help provide constant and intelligent oversight of the spaces and machines in a data centre to improve business outcomes.”
Noteboom, who previously oversaw data centre strategy for Yahoo and later Apple, founded the company, according to a Root position paper issued when the agreement was announced, to make AI available to front line experts.
Sensors, it said, become a “natural extension to the technicians walking the data centre floor, listening, looking and feeling. For example, in the past, an engineer who heard a strange grinding noise coming from the data centre would identify it as a broken bearing and address the issue. With LitBit, the technology will alert the engineers to the noise and they, in turn, will identify it.
“The next time a similar noise surfaces, the system will know what it means and diagnose accordingly. At its core, it’s human insight with an overlay of machine learning. The operational knowledge base grows over time, making the system more and more intelligent.”
According to Byers, reliability and uptime are key considerations for any data centre user and in Root’s case, they range from cloud service providers, hosting companies and video game to large scale IT organizations.
He says that within Root’s 50MW location, at any given time you can have 50 generators around that building and having people to go in and conduct an audible check on what’s going on with our infrastructure during an outage is challenging.
“We started looking at how can we augment our technical team with AI? The first project involved adding an audible listening technology to the entire generator platform and we trained the AI on what a generator should sound like under normal conditions whether it’s sunny out or raining out or windy out. The AI can tell us if there is something wrong with a generator or if it is making a weird sound – ‘check Generator 22 and have someone take a look at it.’
“There are many things we will be able to do with it that will augment the data centre operator walkthrough.”
Root’s challenge, the position paper states, is “one that is industry wide: Given the size and complexity of its facilities, it is impractical for technicians to be everywhere all the time, listening, watching, analyzing and proactively alerting the team to a potential issue.
“During an emergency, such as a utility outage when generators are in use, is when the use of the LitBit system is most important. While there are redundant generators in place to maintain power supply to the data centre, it becomes even more crucial for the operating generators to remain active.”
The company says that an “end-of-period analysis” of the results is expected in December.