AIOps exec bets on incident response market shakeup

AIOps and IT automation have been hot topics in IT for about three years, but the ultimate vision of hands-off incident response has yet to be realized in most IT shops, says Vijay Kurkal, who was appointed CEO of Resolve Systems on Jan. 16. Kurkal had served as chief operating officer for the company since 2018.

Kurkal's key priority in the first quarter of 2020 is the release of Resolve Insights, a platform that folds AIOps IP from the company's August 2019 acquisition of FixStream into its IT automation software. While enterprise IT pros have been slow to trust such systems -- which rely on AI and machine learning data analytics to automate common tasks such as server restarts -- they have begun to find their way into production use at mainstream companies.

Vijay Kurkal

Resolve faces a crowded field of competition that includes vendors with backgrounds in IT monitoring, incident response and data analytics. SearchITOperations.com had a conversation with Kurkal this week about how the company plans to hold its own in this volatile market.

Your product pitch sounds familiar to me. I'm sure I don't have to tell you there are many vendors out there pursuing a vision of proactive IT automation assisted by AI. How will Resolve and FixStream be different?

Vijay Kurkal: There are two ecosystems we're playing with. There are application monitoring tools like AppDynamics, Dynatrace, New Relic, etc. The users that they are going after are the application operations team. FixStream is complimentary to them. But they have limited visibility into hypervisors and deep into the network infrastructure. FixStream builds out a visual topology of every single infrastructure device that a particular application is touching, and all the events are overlaid on that. It's [built] for the IT operation teams that are supporting critical applications.

Some of the other AIOps vendors using AI technologies, they have tons of algorithms, but any algorithm is only as good as the source data. It's a garbage in, garbage out. Our starting point is always around relationship dependency mapping and getting data in context, and prioritizing what to act on. A second differentiator is that AI/ML algorithms are all based on a probabilistic model. [They] say what they believe are the potential root causes [of an issue], but they can't say that with certainty. Where we're taking it is, as soon as those events trigger an alert from FixStream, Resolve automates diagnostics. Typically, that requires a network engineer. We're already trying this out with some pilot customers and by end of Q1 are going to have a product there. Most AIOps companies identify events; they don't resolve them.

Most AIOps companies identify events; they don't resolve them.

Is there a plan for IT automation beyond diagnostics?

Kurkal: The next step, and I don't think most customers are there yet, is, 'I've done this 10 times, and I feel very comfortable, just run this [process] automatically.' You'll have categories of events -- there'll be 30% that are not super critical. As the organization gets comfortable, these can be completely [automated]. Then there are 50% that are critical, and we can give them diagnostics, and point them in the right direction to solve them pretty quickly. Then 10% will be outliers where no automation can help, and that's where IT ops experts will always be very, very relevant to run the business.

Another important aspect of observability is informing the development of the product at the other end of the DevOps pipeline. How does your product work within that process?

Kurkal: The people who build the applications know exactly what stresses their application is putting on various elements [of the infrastructure]. We want to equip the DevOps team with a drag-and-drop system to write automation -- to tell the IT operations team, here's the configuration of the infrastructure I'll need, and here's a set of diagnostic scripts, and remediation automation that's pre-approved. And then it'll be a closed feedback loop where the operations teams can give feedback [to the developer]. We're not saying we'll solve every need of the application, but we are trying to bring together these two teams to drive automation and intelligence.

There are some tools that specifically tie outages or incidents to changes in code -- could Resolve make another acquisition in that space or further build out its products to address that too?

Kurkal: For us, it's a strong possibility in late 2020 or in 2021. It might be an organic development of our products, or potentially, an inorganic acquisition around that. But we do see that's where the market is moving, because no one wants to be reactive, and they want to have it all together.