The rise in demand for next-generation processing power is one of the defining challenges facing data centres today.
It’s one reason why, according to Gartner, 30% of data centres will stop being viable if they don’t implement AI and machine learning technology.
When you consider the advantages, you can see why.
Data can be analysed and managed quicker and at greater scale – which is of course critical for today’s data-fuelled and tech-disrupted world. For example, monitoring and distributing workloads to keep data flowing efficiently.
AI can support data centre operators to add more workloads to physical architecture without fear of downtime. For example, rather than reacting to an outage, AI can monitor hardware performance for warning signs. When it detects potential for an outage, it can automatically take remedial action.
Data centres increasingly rely on AI to identify and tackle the next-generation threats that are becoming a feature of Industry’s 4.0 security landscape. For example, polymorphic malware is able to evade traditional detection methods by dynamically changing its code. Polymorphic malware can also change its behaviour when it senses its being checked by a company’s email protection system. AI-built defences are able to keep up with polymorphic malware in a way that traditional firewalls simply can’t.
There are also great AI opportunities when it comes to cooling. By 2030, the entire IT sector is predicted to consume as much as 20% of the world’s electricity. To reduce usage (and costs), data centres are using a variety of methods to keep data centres cool. For example, locating data centres in the sea, or in the Arctic.
DeepMind in the digital data centre
Google uses AI to manage cooling for its data centres. Its DeepMind technology, acquired in 2014, is tasked with improving energy efficiency and reduce carbon emissions. While human operators oversee its activity, day-to-day decisions are taken by DeepMind. First, a snapshot of a data centre’s cooling system is taken from thousands of sensors. The snapshot is then fed into a machine learning neural network. This then generates predictions of future energy consumption based on actions within the data centre.
For Industry 4.0
Ultimately, there are so many variables which can impact how a data centre runs, too many for a human to be able to calculate and respond to.
AI is ideal to factor in all these possible permutations and come up with the optimum efficiencies. What’s more, as time goes on the AI gets smarter, with machine learning further improving and optimising processes. Naturally, this continuous improvement is what’s needed to compete in the Industry 4.0 era.