Artificial intelligence in the data centre
Though its definition is hotly contested – in particular, its distinction from machine learning – the implication of its success is clear. Take processes and tasks where human judgement is required and replace the human element with a machine that is able to exercise ‘judgement’ on the basis of what has happened before – just like we do.
To make this happen, however, requires huge amounts of computing power. Consider, for instance, the infrastructure required to remotely track vehicle maintenance information, relay that data to a centralised system, carry out data analysis on a program that itself probably sits on the cloud, then allow an artificially intelligent system to make maintenance suggestions or take actions.
Behind all this compute lies the data centre. As the workhorse of the internet and its many billions of connected devices, these facilities fuel much of what we take for granted. Recent years have seen big developments in virtualisation, automation and data centre infrastructure management, but what if AI was applied to the data centre itself?
A smart start for the data centre
It’s already happening, to an extent. Systems such as Cisco’s M5 Unified Computing or HPE’s InfoSight are trying to alleviate the fact that humans are increasingly unable to deal with the complexity of a modern data centre.
Cisco’s system, for instance, allows data centre managers to define usage policies and then allows software to automatically move around resources to put the data centre in the optimal state. Joann Starke, senior manager for hybrid cloud and machine learning at Cisco, notes that “data centres have become so complex that you need to have software in control and automatically — autonomically, as it’s called – making changes in real time.”
According to Schneider Electric, longstanding data centre technology has been using some form of artificial intelligence for a while now. UPSs and cooling units, for instance, use algorithms to decide how the equipment should behave depending on conditions.
Power and cooling equipment often gather data about both the outside environment and its internal processes. In turn, a machine will decide how to respond to this data, for instance, to send a warning message to an operator or to turn off battery charging. Schneider believes, however, that there is a further step to the artificially intelligent data centre.The data centre, but not as we know it
Scott Noteboom, founder of LitBit, developed his AI data centre solution as a result of a very human problem. As data centres moved out of the big cities and towards more remote towns and cities, the initial solution was to pay skilled engineers big salaries in order to entice them away.
However, once Noteboom, who was previously head of data centres at Apple, tried to do the same in a municipal town in China, he found that the expertise simply didn’t exist.
Humans are, unfortunately, not very scalable. This is particularly true when it comes to skilled people – which those working in data centres certainly are. The dream for Noteboom was to simply replicate the skilled and experienced engineers he already had, wherever he wanted.
This is the basis of Dac, his AI persona for data centre management. Dac started off with more than 10,000 pieces of innate knowledge, a resource taken from the collective knowledge pool of many data centre engineers. It can also, being artificially intelligent, learn as it goes.
Like a human, it will also react to its environment, whether that’s noises, the sight of a person, how many people are in a certain area at any one time, and so on. Unlike a human, it can cross-reference this information against an internal bank of information on the correct conditions for the data centre.Will people be banished from the data centre?
George Weiss, vice president and distinguished analyst at Gartner, has argued that IT leaders should be much more ambitious and long-sighted with their plans for the data centre, and recognise the challenges that come with increasing complexity.
“IT leaders need to think bigger in scope and adopt principles of self-organizing systems of intelligence in data centres, edge and clouds of the future,” says Weiss.
“The goal should be to architect platforms and services that monitor and analyse system behaviours, resulting in continually optimized outcomes to predefined goals and service levels.”
For those with access to the coffers who might be wondering what the true advantages of allowing a machine to make decisions might be, researchers at Gartner have argued that doing so can ‘deliver purposeful actions that fulfil the goals set by the organization.’ In other words, say what you want, let the machine do the work, and reap the benefits.
And with research from Accenture showing that AI could potentially boost profitability rates by an average of 38% across all industries and give an economic boost of $14 trillion by 2035, getting AI in your data centre seems like a smart move forward.
There are, of course, always fears about job loss in any discussion about AI. But the same fears always arise whenever a game-changing new technology appears – from the Luddites of the Industrial Revolution to the cloud naysayers. But with it always comes new job creation, until eventually, the technology becomes the new normal.