Why the potential of AIOps needs no hype
- 07 March, 2019 02:00
It’s hard to separate the hype from the reality when it comes to artificial intelligence (AI). Once the domain of books and movies, it’s now being rapidly applied to all areas in business. AI for IT operations, or AIOps, is no different.
AIOps combines machine learning and big data to enhance IT operations by automatically spotting issues and in some cases fixing them in real time. It’s easy to see why enthusiasm has been coming in quickly, given the potential for AIOps to support a business’s need for speed and agility.
CIOs are being pushed to find ways to streamline their operations, and free up money and capital for innovation and AIOps. That’s why many CIOs say AIOps will be central to their plans in the coming years. A Gartner report found 40 per cent of large enterprises will be using AIOps by 2022, up from 5 per cent today.
While it’s early days for AIOps, you’ve probably already heard all kinds of predictions about the revolution it will launch. With just a few quick-and-dirty algorithms to work from, some commentators are promising AIOps will analyse your worldwide IT operations and tell you exactly how many data centres to shut down, to save tens of millions of dollars.
But hype like this is dangerous. It leads to disillusionment that can stop real tech advancement from ever getting off the ground. Companies abandon technologies that don’t live up to overblown promises.
In the case of AIOps, that would be too bad, because it really will have a staggering impact on our world. It can automate mundane daily tasks, help resolve problems faster and, more importantly, look ahead to detect and fix issues problems before they occur. By extension, it will improve customer experience and free your people to work on more important matters, like innovation.
In fact, it already has in major companies like Google. Unsurprisingly, the tech giant has a vast need for electric power in its data centres around the globe. The bill for cooling alone costs the company millions of dollars a year.
Data centres generate millions of data points a day from thousands of sensors, yet that data has typically only been used for monitoring. So Google put AIOps to work, mining the data it already had about temperature, power usage and cooling, and using machine learning to figure out ways to cut cooling costs. The result was huge: a 40 per cent reduction in the power required to cool data centres, and a 15 per cent fall in the amount of power used in its centres. Those are real-life savings, not pie-in-the-sky promises.
And it’s not the only example. The power of machine learning applied to big data can be put to work by IT Ops teams every day to drive benefits. AIOps provides insights that can significantly reduce the cost of operations.
Take contact centre platform business, NICE inContact. It was spending millions on IT infrastructure to sustain its fast-growing business providing customer call centre software globally. By employing AIOps, machine learning and advanced analytics, the company was able to correlate IT infrastructure usage data with business growth projections to determine what IT resources to buy and when. With these insights, it eliminated IT capital purchases for five consecutive quarters and was able to plan effectively for future needs. This provided significant savings while ensuring the company could deliver the service that over 200,000 call centre agents rely on daily.
At any company, AIOps can provide insights and automate actions to manage straightforward everyday tasks that take up IT’s time and money. That can include automatically responding to an alert that a server disk is almost full, freeing up disk space by deleting log files. Or predicting a performance issue, pinpointing the problem and then automatically remediating. Instead of spending hours determining the root cause of a performance problem and more time recovering from it, enterprises can detect and address problems quickly - or even avert them altogether.
This is the potential of AIOps. By handling IT’s everyday problems with actionable insights from machine learning and automation, it can avert problems, cut costs, improve customer experience and free IT staff to work on the innovations your company desperately needs. So forget the hype about AIOps doing near-magical things - the reality is good enough on its own.
Stuart Fleetwood is Asia Pacific director, digital service operations – analytics, security, automation, at BMC Software.