Is it possible to automate all of cloud operations?

As automation gets better and humans withdraw from the process, we could achieve almost 100% automation of cloudops and secops in just a few short years.

Is it possible to automate all of cloud operations?
Putilich / Getty Images

As I move from project to project, I’ve seen the latest trend is to leverage operational tools, such as AIops and security operations platforms to automate most of what it takes to proactively operate a cloud, hybrid cloud, or multicloud deployment. This means automating everything from routine management and monitoring to shutting down and starting servers to work around problems, and all the while machine learning on the job (that’s the AI in AIops).

Nobody is ready to retrain their ops staff yet, but it’s clear that advances in root-cause diagnostics and self-healing processes, business continuity and disaster recovery, and other services that make up the daily life of a cloudops engineer can be automated to be more reliable than humans. We’re now dealing with tools that can learn, that improve as they experience operations, that can perhaps work better than a human, eventually.

Automation of cloudops is much like the automation of driving. Although we know that the technology can drive the car—perhaps better than we can—the idea is still intimidating. Cloudops automation is much more complex than driving a car, but many of the same types of problems will have to be overcome. The result is a set of automated processes that may result in a much better-run cloud, as well as a proactively aware security system that becomes better over time.

Will we take the leap? Take our hands off the steering wheel?

My view on emerging technology is that it takes about three or four years from the time the technology is capable to the time that it’s widely employed. Considering that most enterprises have taken 10 years to get just 20% to 30% of their workloads into public clouds, this may be a longer trod than we think.

A critical success factor will be for cloudops and secops automation to provide much better results than traditional approaches, meaning humans. I figure a few aggressive upstarts will be the first to go hands-off; once they prove successful, others will follow.

It’s been this way as long as I’ve been in the technology business. Even though technology is all about leaps of faith, everyone wants someone else to leap first.

Effective automation will be a true force multiplier for a business. Automated ops processes can scale and become much more advanced and effective over time. The scary part is that if humans took back control for whatever reason, they most likely would not be as effective as the AI-powered automated ops processes.

This is not science fiction; this stuff is working today. Were it not for the possibility of getting walked to the edge of the property for doing so, I could set up deep automation for 80% of the manual processes for most enterprise cloud deployments, moving to 100% in just two years.

The trick is dealing with the expectations and fears of humans before we can replace humans. Now, is that not a conflict of interest?

Copyright © 2021 IDG Communications, Inc.