2017 was the year in which it became obvious and undeniable that software has already eaten huge chunks of the world. This realisation means that to be competitive, it is no longer sufficient simply to add some software; instead, every single part of the organisation must be augmented through software.
At the recent “Gartner IT Infrastructure, Operations Management & Data Center Summit”, Dave Cappuccio said, “If you create a five-year plan, you’re chasing a ghost”. This standard planning window is not only no longer useful, but is actively harmful. The rate of change, in terms of both the technology itself, and the demands that the business places on IT to deliver and operate that technology, is so rapid that any plan will be hopelessly obsolete by the time five years are up.
This acceleration in the rate of change has made it more and more difficult to look into the future and make predictions. The trend has been developing for a while, as I wrote after last year’s event. In fact, even science fiction writers have given up. William Gibson, who famously invented cyberspace in his debut novel “Neuromancer”, set each book of his first trilogy of this century in the year prior to its own publication, precisely to avoid having to make forecasts in this unpredictable situation.
Given all of that, it seems foolhardy to try to look too far ahead, and especially in a field such as AIOps, which is where some of the most rapid motion is to be found. However, there are a few macro-trends that are still worth looking at. With any luck, enough of this should hold up that it won’t look completely off-base by this time next year - although there might still be the odd dot-matrix printer spewing fan-fold paper in a space yacht somewhere.
The rapid evolution of the field of AIOps is what complicates any prediction, but is itself a prediction. As the market, meaning the combination of vendors and users, starts to work out collectively what AIOps is and should be, a certain amount of welcome clarification should be expected. The same trend can be seen in past waves of technology adoption, such as cloud computing. Part of the very definition of AIOps is that it crosses the boundaries of various other disciplines (monitoring, service desk, and automation), and so it does not have its own Hype Cycle. However, it is clear that it is still in an early upswing phase when anything is claimed to be part of the new wave. At this point on the cloud hype cycle, we were hearing claims such as “mainframes are really cloud computing, because users rent capacity instead of buying infrastructure”.
Over the next year, we can expect it to become clearer what is AIOps (and what it is not). This will help practitioners determine how AIOps can fit into their existing strategies and processes, helping them to assemble their IT Operations toolchain to best support their users and customers.
A corollary to that clarification of the definition of AIOps will be a certain amount of consolidation. Expect to see both legacy vendors hoping to inject some innovation into their own existing suites by acquiring technology from outside, and different approaches to AIOps being brought together under one roof to deliver a more holistic solution. In this phase it is always crucial for prospective buyers to do their homework and read the small print, to make sure that the shininess is thicker than just a layer of paint over some poorly-matched parts.
On a more positive note, 2018 will be the year when we will start to have meaningful volumes of stories from early adopters of AIOps about both successes and lessons learned from their initial projects. Some companies have even had the opportunity to go around a couple of iterations, applying insights from one AIOps-enabled project to the next, and reinvesting the dividends of success in new improvements. This will in turn give more confidence to those who had been waiting on the sidelines, intrigued by AIOps but uncertain about the true requirements and expected benefits. By the end of this year, AIOps can be expected to be at least a mainstream topic of conversation, if not yet universally adopted.
A big part of AIOps is integration, both between technical components and between the different teams that use them. As adoption increases, this integration can be expected to deepen, as pilot projects broaden their base beyond the initial few teams, and new teams get involved, bringing their own tools with them. This extension to the overall IT Ops toolchain will include commercial tools, open-source components, and home-grown utilities that are already in use in companies and organisations today. Bringing these tools together, enabling frictionless flow of information and actions between them, will go a long way to breaking down the silos that have been slowing down enterprise IT.
2018 is the year in which AIOps becomes part of a mainstream conversation in enterprise IT. The biggest transition is that individual projects will no longer be considered in isolation, but rather as part of an integrated whole. AIOps is one of the key enabling factors for that process of tighter integration, and 2018 is when that will become more visible and widely accepted.