Q&A: Why AIOps will dominate IT operations

Prepare to welcome new AI masters with the integration of CA's machine learning into DevOps automation tools, Automic's CTO says.

The CTO Automic Software said that AIOps will be the focus for DevOps automation tools, as well as for IT operations careers.

John Purrier, CTO of Automic since late 2015, helped steer the company toward an IPO until it was bought by CA Technologies for $635 million last December. The two companies have finished integrating their overlapping release and workload automation products.

So what's next for the newly merged companies? Gartner's newly minted term is AIOps: the addition of machine learning and artificial intelligence (AI) to DevOps automation tools. caught up with Purrier to learn how this concept will influence the development of CA's products and affect IT operations jobs in the future.

Who are CA and Automic's joint customers? What types of companies are you talking to most?

John Purrier: The ones I've been involved with have been big banks. There's overlap in terms of joint customers in the financial sector and retail as well. There have been some really interesting conversations with some of the global service providers, and being able to use Automic to create platforms to deliver CA products through their efforts.

What's happened since the acquisition closed?

Purrier: So far we've been working closely with the other business units within CA. They're very excited to have that foundational automation piece in the portfolio. That was the missing piece of their story and we are able to work with the continuous delivery business units, the mainframe guys, security guys and so on, and bring a set of capabilities to CA that makes everything much more valuable. So if you're using the automation engine either for workload or release automation, we now tie into things like Test Data Management, BlazeMeter and so on.

What's on the roadmap for the companies this year?

Purrier: Our timeline is to release version 12.1 at the end of September again like we did with version 12. We something called environment provisioning in our release automation product in version 12. The initial system that we put in place was all about Docker containers. You can set up profiles or blueprints and then those will provision, deploy, run and deprovision containers on demand. In version 12.1 we're expanding that into VMware, .NET systems, and adding support for things like container schedulers such as Kubernetes.

How about beyond 2017?

A lot of [the AI underpinnings] will be behind the curtain -- what we need to do is expose the knobs and levers that will allow a human being to optimize the system.
John PurrierCTO, Automic Software

Purrier: We have efforts going on further out on data transport, collection, storage and analytics. CA also has a data science team and I'm engaging them to figure out how we model automation systems mathematically so we can start to use this data to drive automation and learning systems. We have two projects in flight -- one is the base data collection, transport and analytics, and the second is modeling automation systems. Very quickly, heuristically driven automation is going to become normal. Using AI and various machine learning systems is going to be valuable. We'll start to this technology in 12.1 from a basic standpoint how you get the data to be able to analyze it. In 12.2 we'll start rolling the automation in. One the benefits to Automic from the acquisition is that CA has people already doing work in this field so we can just take advantage that rather than having to figure it out ourselves.

Will heuristic automation make IT ops professionals extinct? What other roles will they occupy in IT?

Purrier: People have the idea that automation means job deletion. A lot of jobs are done by human beings today that don't need to be done by humans. Those positions will eventually get displaced using these sorts of intelligent automation systems. It doesn't mean that people go away -- it means they get shifted to higher-value positions. It's a way of upleveling the value that IT and IT ops bring to the business.

When I've walked through these scenarios with the CTOs of big companies, they're in agreement. Usually when we start a conversation about DevOps, the goal is budget reduction and you can save money through automation -- but ultimately we can convince them you can only reduce a budget so far. But if you create value, the upside is infinite. The real value of automation is in value creation, not necessarily budget reduction. Of course, if there's the ability to do short-term budget reduction while the DevOps transformation is happening, that's a win for the executives we're talking to as well, but ultimately they want IT and IT ops to be a value driver for the business. That's where we steer people.

Can you give me an example of where IT ops can add value?

Purrier: Instead of having folks that sit in front of screens looking for something to go from green to red, we let the system take care of that. What the system won't be able to do out of the gate is ... [create] the model for automation and analytics; human beings are going to be needed to put together patterns. The value is in tweaking the model. Ultimately, if I was an IT professional I'd start steering that way. That's where human beings are going to outpace computers for a long time. Then you update your machine learning systems -- they're just automation for the maintenance and optimization of the models. There's a nice symbiosis there.

How can people obtain these skills?

Purrier: People don't necessarily need to understand the AI underpinnings -- we need to create tools that expose this stuff in a logical way. At this stage, it's on us as the vendor to create user experience models that allow people to pick up the tooling, and learn that as a school rather than going out and learning how TensorFlow or Microsoft Machine Learning systems work. The stuff I'm talking about will be delivered as a service, and a lot of it will be behind the curtain -- what we need to do is expose the knobs and levers that will allow a human being to optimize the system for their own particular situation. It is a challenge, though, even looking at setting up a service -- finding people today that are experts in this field is tough.

Beth Pariseau is senior news writer for TechTarget's Data Center and Virtualization Media Group. Write to her at [email protected] or follow @PariseauTT on Twitter.

Next Steps

AI is a hit -- but the World Economic Forum is leery

App support tools benefit from automation

Software-defined networks prime data centers for automation

Dig Deeper on Configuration Management and DevOps