DevOps is dead, but what will replace it?
I’m going to make a bold claim today and say that DevOps is dead; the amount of time spent doing DevOps will be reduced to less than 20% of what it is today within the next five years. And all this while the use of cloud and edge computing continues to grow exponentially. How can this be possible? With a simple word that a few embrace, but many fear: Automation. Perhaps that sounds redundant to you? Automation, after all, is one of the main tasks involved in DevOps. But in this case, developers might be coding themselves out of a job…
Now before you start panicking, let me reassure you; it’s all going to be okay. This is not a doomsday scenario that makes unemployment skyrocket and leaves the economy in the dust, quite the opposite. It will make companies more efficient, more productive and free developers time to work on things that really matter: building great products. What I’m referring to here is mostly the cloud-related aspect of DevOps, which is poised to undergo a massive change due to the automation of its main tasks in the next few years. Before diving a little deeper, a little word of caution. DevOps has grown to mean many things to different people. Initially, it was meant as a development philosophy to deliver and improve a product faster by removing the “development” and “operations” from being siloed. Today, a quick search on any jobs site will give you a myriad of “DevOps engineer” roles, where the main tasks involve very little product development. The distinction matters little in our hypothesis; time spent doing DevOps will be significantly reduced in the next few years due to technological advancements.
Do we have the tools to do this?We do not need an AI with the capacity of Skynet to automate the cloud management part of DevOps; it is possible to do it today. But you need a specific set of tools to make it possible.
An easily distributable cloud application
A highly scalable orchestration platformA multi-provider, multi-network approach to ensure reliability
A set of mechanisms for the auto-regulation of any issues
It sounds almost too simple, why did no-one think of this before? Well, most of the pieces had been missing until not long ago. You can’t get the multi-cloud before you have a strong cloud ecosystem. It was hard to easily distribute applications on the cloud before containerization. It is still very hard to make a completely auto-regulated and highly scalable system. But the wheels are turning and the train has already left the station…
Automation comes with a price tag, who’s going to pay for this?
It will pay for itself. There is a multitude of financial incentives today that make automation a wise economic decision. Let me list a few out of the top of my head:
Having internal developers do DevOps work is inefficient; time would be better spent working directly on a product
Hiring an external team for handling DevOps is inefficient and expensive
Engineers with DevOps experience are one of the most sought-after resources in the IT world, making it hard to find the good ones, and expensive to hire
Failure is costly in our fast-paced world where everyone expects things to work perfectly at all times
Automation can do the work of many engineers for a fraction of the price
Automation can find ways to improve upon its own system in ways that humans cannot compete with
I could go on, but I think you get the point: most companies will be thrilled to concentrate their efforts and dollars on other endeavours. Talented developers will be able to go back to building great things instead of being stuck managing infrastructure and monitoring the network.
When is this train going to steamroll into my industry?It’s already happening. With the right tools, it’s now easy to distribute micro-services and serverless functions, and they are just the beginning. The day where a developer will simply upload a cloud application and let an automated system take care of its distribution, scalability, reliability and auto-regulation is upon us. That means no more logging manually into virtual machines, no more servers provisioning, no more figuring out the best cloud and best location to host an application.
The list of time wasters goes on and on.I, for one, welcome our new automation overlords.