Is AIOps NoOps? No, But It’s the Closest We’ll Come

Making IT operations simpler – which AIOps does by helping teams to make smarter, more informed decisions about complex monitoring and APM problems – is great.

But what would be even greater is eliminating the need for IT teams to make decisions at all – a prospect known as NoOps. By automating application management to the point that human involvement is no longer necessary, NoOps offers tantalizing possibilities for the IT operations teams of the future.

Will organizations eventually reach a state of NoOps by making greater and greater use of AIOps? Or is it true NoOps simply unachievable? Those are impossible questions to answer without the ability to see into the future, but they’re nonetheless worth discussing in order to gain perspective on where AIOps stands today and how it might evolve in the future.

What Is NoOps?

Short for “no operations,” NoOps is the idea that IT operations may become so completely automated using AI and ML that human IT operations engineers no longer have to manage systems manually at all. They can sit back and let automation tools do their work – or, perhaps, the IT operations team would simply cease to exist at all, because its job would be fully automated.

The NoOps concept originated in 2011, when Mike Gualtieri of Forrester Research wrote,
“I Don’t Want DevOps. I Want NoOps.” His vision was simple enough: “Using cloud infrastructure-as-a-service and
platform-as-a-service to get the resources they need when they need them,” Gualtieri wrote, developers would someday be able to achieve everything they needed without having to collaborate with IT operations teams at all.

NoOps and AIOps

Notably, Gualtieri introduced the NoOps concept well before the emergence of the idea of AIOps, a term that was coined by Gartner in 2016 as shorthand for “AI for IT operations.” When Gualtieri was thinking about eliminating IT operations, he mostly envisioned cloud services, not AIOps monitoring and management tools, as the key to doing so.

Nonetheless, you could argue that AIOps does much more to bring the idea of NoOps closer to reality than cloud services ever will. IaaS and PaaS are great, but you still need someone to manage them. Even if you automate infrastructure rollout and application deployment using IaC, someone still has to write the templates, make sure they are deployed properly, and update them when requirements change.

In contrast, AIOps opens up the possibility that complex IT operations tasks can be entirely automated. AIOps tools could automatically detect and remediate an application performance problem by allocating more resources to the application, for example. Or, they could automate capacity management by scaling infrastructure up and down as requirements change. These are tasks that cloud services typically can’t handle. An IaaS service isn’t going to tell you exactly how many nodes you need in your cluster, for instance. A PaaS service might automate application deployment, but it won’t automatically fix an app that crashes because of an unexpected environment condition.

It’s worth noting, too, that the ability of AIOps to eliminate operations work extends beyond IT operations narrowly defined. Some aspects of security operations and DevSecOps can also be automated using the approaches that AIOps enables. AIOps tools could automatically update firewall rules to manage network security threats, for example, or they could automatically isolate endpoints that they determine to pose a risk. These are also tasks that cloud services can’t automate.

The Limitations of AIOps for NoOps

The above notwithstanding, it’s important to maintain a healthy perspective on the limitations of AIOps and the extent to which it can enable true NoOps.

AIOps tools continue to evolve. They are getting better and better at not just detecting problems, but also automatically remediating them. They are also becoming more adept at addressing domain-agnostic needs, which means they can handle applications in any type of environment, rather than being limited to working in only certain types of domains.

Nonetheless, it remains difficult to envision a world in which all aspects of operations are totally automated. Unforeseen issues will always arise, and even the smartest AI won’t be able to remediate all of them on its own.

More Efficient Ops

IT operations teams, then, probably need not worry that their jobs will disappear anytime soon due to AIOps tools. IT engineers will always be necessary to solve the truly challenging problems that AIOps can’t.

But IT teams can look forward to a future in which more and more aspects of their jobs are automated. Rather than posing a threat to IT, this future will enable IT engineers to focus on more interesting work. Instead of manually interpreting log data or increasing infrastructure capacity, IT teams can let AIOps do the boring work while they spend their time helping developers plan application enhancements, assessing new deployment technologies, optimizing cloud architectures, and so on.

Just as important, AIOps will also help IT teams manage larger-scale environments without a proportionate increase in team size. That advantage will be critical as applications become ever larger and more complex.

In short, then, AIOps is poised to make IT operations more efficient, but probably not to eliminate it entirely.

Conclusion: NoOps for the Real World

Ultimately, NoOps is an unrealistic dream. I don’t think that even Gualtieri actually expects NoOps ever to become a reality; he seems to see the concept as interesting food for thought rather than a realistic goal.

What is totally realistic, however, is AIOps. Today, AIOps has ushered in a level of efficiency and automation for IT operations that even Gualtieri could not have imagined when he first wrote about NoOps a decade ago. And AIOps will only get better over time, even if we never reach the point of NoOps.