This content is part of the Conference Coverage: Guide to DevOps Enterprise Summit 2017

New Applitools, products make implementing DevOps easier

The DevOps journey is well underway, but many obstacles remain. New products using artificial intelligence and machine learning are trying to make DevOps implementation simpler.

It's no surprise implementing DevOps is tricky. What is surprising is the growing number of companies turning to artificial intelligence to help make things go more smoothly.

Just in the last few weeks, startup began offering a continuous delivery as a service platform for developers that uses machine learning to monitor an application and roll it back as necessary. And another new company, Applitools, is also offering an AI-powered, cloud-based approach for test automation.

In other words, it's the bots to the rescue.

Torsten Volk, managing research director for hybrid cloud, software-designed data center, machine learning and cognitive computing at Enterprise Management Associates, based in Boulder, Colo., said when implementing DevOps, the hurdles to an average developer pushing out code are quite challenging, and using AI and machine learning could simplify things dramatically. Volk, who will speak at the DevOps Enterprise Summit 2017 in San Francisco on DevOps in the enterprise, said it's time to streamline the process -- something he calls "intent-based DevOps."

"If a developer has a task, a developer should receive a workspace that is dedicated to that task," Volk explained. "Right now, a developer doesn't have a realistic environment where he can test something or try to experiment with scalability. You want to put the intent of that application into the workspace the developer is using to speed up the process."

That way of implementing DevOps means using machine learning, which can be quickly trained in an organization's best practices, to speed things along and give the developer the necessary help and tools. “If you can do that, you eliminate the worries,” Volk said.

Automation and implementing DevOps

That's certainly the intent with Harness, which in fact was named for the concept of a safety harness, said founder and CEO Jyoti Bansal. Harness is aimed at an area nearly every company struggles with when implementing DevOps: automation.

"There are two problems that happen when people are trying to automate," Bansal explained. "The first is how do you automate. You create tons and tons of scripts, but they're unmanageable and not maintainable. And the second part is, once you do something, the code is so complicated, it's hard to verify. The goal of Harness is to solve those two problems: Make automation easy and verification really, really easy."

Harness tackles this problem by first allowing automation scripts to be created using a simple drag-and-drop GUI. Bansal claimed this approach takes a three- to six-month scripting job and whittles it down to 30 minutes. Once the automation is complete, when implementing DevOps, it's vital to continue to ensure it's working, and that's where the AI piece comes in.

"We use machine learning that knows what is normal, so it knows when something is not working," he said. "Are we seeing more errors or deviations? We're building that safety net in, so developers can move as fast as they want to. And if they fail, Harness is there to catch them."

Applitools for implementing DevOps

AI is also the cornerstone of the Applitools offering, which was designed to make visual testing so fast it could be a reliable piece of a continuous delivery effort when implementing DevOps, said Gil Sever, co-founder and CEO. "At a high level, we make sure that an app looks right on different devices," he said.

At the heart of the process is a new AI-powered technology that can break down a computer or mobile screen just like the human eye does, Sever said, and can then very rapidly spot any problems. "And if it finds a problem, it can decide, like a human, if that's something an actual human would notice or not. And if not, it doesn't flag it," Sever said.

When implementing DevOps, Applitools can be used for production testing tasks, including regression testing and cross-browser testing. And it can even be used to monitor transactions. "We're able to compare the number of steps done visually with what we've seen on hundreds of other apps," Sever said. "If there is something unusual, like it takes more steps than usual or is done in a different way, it will signal the team that made the app so they know there is something different and less efficient."

According to Volk, the power of AI really does come down to efficiency. "People need to see that it sounds like science fiction, but most of it can be done today. The reason why it's not done is that everyone is so bogged down with keeping the lights on."

Next Steps

Why AI and the cloud are a perfect match

Be smart about when and how to use AI

Will DevOps really do the job?

Dig Deeper on Agile, DevOps and software development methodologies