In a sense, the holy grail of DevOps is to achieve what’s known as NoOps—a state in which IT operations are so completely automated that they demand no effort on the part of engineers at all.
Traditionally, total NoOps has seemed a fantasy. Few people have thought that IT operations of all types could be completely automated.
But with the help of AIOps, such an idea is now less fantastical. By leveraging new techniques as the basis of automation, AIOps can automate IT workflows that were previously impossible to manage without manual effort.
Here’s a look at what NoOps means, how AIOps helps make it practical, and how much has already been achieved in the march toward NoOps.
What Is NoOps?
The goal of DevOps is to make IT operations teams and developers work more efficiently together.
NoOps, meanwhile, takes this idea to the extreme (or negates it, depending on your point of view) by proposing that IT operations—or the human part of them, at least—can be eliminated altogether.
In a NoOps environment, IT operations like deploying applications, monitoring performance, and addressing security problems would all be performed by software using automation tools. Human engineers would not have to worry about these tasks, and would be free to focus all of their efforts on creating and enhancing applications themselves.
The (Traditional) Limitations of NoOps
When Forrester first proposed the idea of NoOps back in 2011, the vision was to leverage cloud-based PaaS and IaaS platforms in ways that eliminated the need to perform basic infrastructure provisioning and application deployment tasks.
There was clear value in automating these workflows. At the time, however, discussions of NoOps were limited primarily to these areas.
In other words, few people actually thought that NoOps could completely, totally automate all aspects of IT operations. The idea was that common IT operations tasks, like setting up hosting environments and deploying new versions of applications, could be automated (which, indeed, they largely are today). There was a sense that more complex tasks, like those related to troubleshooting unexpected problems within software environments, would always have to be handled manually.
You could say, then, that NoOps has traditionally been seen as a realistic goal only if you recognized that it couldn’t be taken to the extreme. Although the term has been tossed around a lot over the last decade, it was not really with the expectation that IT operations would actually disappear entirely. It functioned more as a provocative way to emphasize the importance of automation, with the unspoken implication that not everything in IT operations can be fully automated.
Realizing the Full Potential of NoOps with AIOps
I’d suggest, however, that AIOps is pushing the IT ecosystem toward a frontier where the near total disappearance of IT operations is becoming realistic.
AIOps, a term coined by Gartner in 2016, refers to the use of AI to automate IT operations. Thus, whereas early proponents of NoOps saw cloud-based toolchains as the key to IT automation, AIOps leverages data analytics and machine learning to automate IT workflows.
Based on this difference, it is unsurprising that AIOps can achieve automation in areas where traditional approaches to NoOps fall short. A PaaS or an IaaS platform can only automate the things that it is specifically designed to automate (which are usually application deployment and infrastructure deployment, respectively). But AIOps can use AI to automate virtually any type of IT task, whether it’s determining how many resources to allocate to a virtual server, load-balancing traffic between redundant application instances, or determining the root cause of an application failure.
Granted, there are limitations to what AIOps can currently achieve. In many instances where AIOps is used today, the automated actions that it takes are based on predefined playbooks or workflows that have to be set up manually. An AIOps-powered platform that fixes an application performance problem by allocating more memory to a server, or closes a security vulnerability by blocking an offending host, is relying in part on guidelines that humans set up ahead of time.
That said, as AIOps platforms become more sophisticated, the need for manually generated guidelines will decrease. The best AIOps platforms will be ones that can use data analytics and machine learning to determine how to identify, understand, and react to problems on the basis of what has worked in the past, rather than on playbooks.
To put this another way, you could say that AIOps in its early form centered on correlating data sets of limited size with preset response plans. Going forward, however, a new generation of AIOps platforms are leveraging machine learning to understand unified data sets and make autonomous decisions that take into account the full and unique context of every situation.
That’s what complete NoOps is all about. Much more than simply automating certain workflows using preconfigured tools, a total NoOps approach is one where decisions can be tailored to each problem in a unique, customized way.
This isn’t to say that we should expect IT operations to disappear entirely anytime soon. There will likely always be some very complex issues that AIOps platforms can’t resolve on their own.
Still, thanks to AIOps, NoOps is starting to look less like a provocation and more like a realistic goal for everyday IT teams. That’s a huge shift from the state of NoOps a decade ago.