Data centre automation & the ‘rise of the robots’
Tue 3 Apr 2018 | Leo Craig
With the shift towards artificial intelligence (AI) and automation on a seemingly unstoppable path, Riello UPS General Manager Leo Craig asks whether the rise of the robots is inevitable, and if so, what the consequences are for data centres of the future.
Nearly a third of the UK’s population owns at least five connected devices. That’s around 20 million people with 100 million-plus smartphones, wearables, virtual assistants and more. But that’s just the tip of the iceberg. Experts predict the Internet of Things (IoT) will see more than 50 billion appliances connected globally by 2020, with the advent of 5G wireless within the next decade helping to create a market worth $7 trillion a year, according to the respected International Data Corporation.
There are already more connected devices on the planet than there are people, which means one thing – data and lots of it! To keep up with this ever-increasing demand, modern data centres are being tasked with doing much more with less. Less space, less energy, and ideally less manpower.
The only conceivable way to satisfy these challenging requirements is to embrace the technological revolution known as Industry 4.0. Automation, big data, artificial intelligence, machine learning – the smart data centre is the only show in town.
So the machines are coming… But are they set to seize complete control?
What is data centre automation?
To put it simply, it’s the automating of tasks that have traditionally been performed by humans, ranging from management and monitoring through to maintenance.
Machines also don’t require sleep, they don’t take holidays, and they don’t call in sick – they’re there 24/7
Or to put it even more bluntly – take tasks which require an element of human judgement and replace the human with a machine that has the capability to ‘learn’ and make similar decisions by analysing information exchanged through sensors on connectable devices. Predicting future outcomes based on data from the past.
The modern data centre has become so large and so complex that it’s virtually impossible for humans to keep up with the volumes and variety of information.
Machines, however, can analyse and process data almost as quickly as they receive it. With smart, connected machines that means real-time, instantaneous responses. It offers data centre managers the flexibility to immediately scale-up as and when the need arises. And it guarantees a rapid reaction to many faults without the need for human intervention.
Of course, machines also don’t require sleep, they don’t take holidays, and they don’t call in sick – they’re there 24/7.
Machine-to-machine communication provides the potential for AI technology to automate many of the functions carried out by IT administrators and ops teams. Reams of raw data rapidly transformed into actionable information.
Used in conjunction with Data Centre Infrastructure Management (DCIM) software, AI can automate many critical datacentre functions, even helping to isolate security threats by remotely severing connections to other systems.
It can take on time-consuming patch management duties, updating systems to the latest, most secure software. It can be used to take on everything from essential tasks such as triggering disaster recovery processes, through to the relatively mundane yet essential requirement to automatically generating end-of-day reports.
Of course, this sort of smart communication isn’t entirely new to a data centre environment, with an uninterruptible power supply (UPS) unit a clear example of this. Sensors on a UPS gather and exchange data about both the system’s performance, such as remaining battery time, and the external environment.
This data is monitored, either on-site or remotely, and analysed to enhance performance, or where necessary, automatically trigger emergency response or system shutdown scripts.
Modern UPS units can also communicate and integrate with smart grids, which opens up opportunities for energy storage, intelligent energy use, and demand-side response. DSR, as it is commonly known, is a concept where energy users are provided with incentives to reduce usage at peak times, helping the grid balance supply and demand without the need for any additional power generation.
In practice, data centres can use their UPS to generate and store an energy surplus during off-peak periods when the per unit cost is lower, which can either be used in peak periods or power outages, or sold back to the National Grid on demand. The UPS does need to be fitted with Lithium-Ion batteries to fulfil this role as a “virtual power plant”, but doing so can provide significant benefits in terms of hitting corporate social responsibility and environmental targets, and even providing an extra revenue stream.
So, what are the advantages of embracing automation in the data centre, apart from the obvious freeing up of IT team time?
The obvious benefit is speed and guaranteed predictable performance with minimal need for human interference. Potential problems can be detected and solved in real-time, promoting uptime and limiting damaging downtime. Another area with huge potential is improving efficiency and reducing data centre energy consumption.
AI in action – Google harnesses its DeepMind
One of the most striking examples of the incredible impact that embracing automation in the data centre can lead to is the case of Google. Back in 2014, the tech giant applied the machine learning of its then recently-acquired DeepMind AI unit to help manage power usage across its sprawling network of data centres.
Analysing historical data from 120 variables such as power usage, pump speeds, and temperatures, the algorithms were able to calculate and implement far more efficient ways of cooling data centre equipment.
The result? A 40% decrease in cooling requirements and an overall 15% cut in energy consumption, substantial reductions that saved the company hundreds of millions of dollars.
While most other organisations obviously don’t have anywhere near the resources – or perhaps even the technical expertise – of Google, adopting a similar thought-process would certainly make a significant mark on data centres, whatever their size.
Indeed, a modern, connected UPS can play an integral role in the sort of smart power monitoring and usage described above. As well as the unit’s sensors continually feeding through vital performance statistics, depending on what the data says, the UPS can also decide to switch into Eco saving mode where appropriate, which results in performance efficiency up to 99%, based on analysing the quality of the mains supply. All these steps help data centres in their search for energy savings and cuts in carbon emissions.
A Brave New World?
In general, the automation of processes in data centres has tended to be used to provide information, which humans have then interpreted and put into action – people are still very much in control. As AI and machine learning becomes more and more advanced though, this separation of powers will naturally become increasingly blurred.
Whereas in the past data centres were seen simply as a means of storage, the smart data centre of the future will analyse, interpret, and process that information in real-time.
Whenever such new technologies or radical processes are introduced, there are always concerns over potential job losses where machines take over tasks previously performed by people. Such fears were voiced by the Luddites and have been echoed in every industrial revolution since.
Taking history as our guide, the outcome tends to see certain roles being replaced, rather than people actually losing their jobs, and that’s the likelihood in smart data centres too.
Compared to many industries, the number of people required to run a data centre is already fairly limited. Increased AI will potentially lead to fewer specialist IT positions with a shift towards more multi-skilled or flexible roles.
Rather than fear these changes, automation should be seen by data centre managers and administrations as an additional tool in their armoury. IT professionals will be able to focus their attention on added-value tasks, rather than simply “keeping the engine running and the car on the road”, to use a driving analogy.
However, no matter how good AI and machines are, at the end of the day, they are only as good as their algorithms and programming. They’re vulnerable to manipulation by people or even other machines, while there are the obvious concerns around data privacy and security.
While AI, automation, and machine learning will certainly play an increasingly influential role in all aspects of our lives, we’re still some way from a world – or a data centre for that matter – being completely run by robots.
Next time we’ll explore how the uninterruptible power supply (UPS) industry is ready to help data centres make the leap forward into a smart, automated future.
Our free Power Continuity For Industry 4.0 whitepaper will explain how modern UPS units are in many ways already ‘smart’ ready and able to assist your transition to becoming a smart data centre.