The Evolution of Observability in AIOps with BMC’s Joseph George

This is techron tv. Hey guys, thanks for the throwaway here with Joseph George, who's vice president of product management in charge of AI ops for the BMC division that focuses on digital services and operations, and we're talking about observability and hopefully how it's gonna get easier in the age of ai. Hey, Joseph, welcome to show.

Thank you, Michael. Pleasure to talk to you this afternoon. I think everybody kind of likes the idea of observability, but for the most part, we're all maybe still doing a whole lot of monitoring and not much observability.

And part of the issue has just been that we can collect the data, but making sense of it is hard and there's too many conflicting alerts. So I think we're about to see some progress with the rise of machine learning algorithms, but where are we on this observability journey? Yeah, That's, that's a really great question, Mike.

Michael, uh, it's interesting. Some folks may even think about it as Observably just being a different name for monitoring. Uh, I've been in the space many years, and I can tell you that's not true.

Uh, ultimately the problems we're trying to solve are really pretty much the same, right? From a business perspective, you wanna make sure your systems are up and running. You wanna make sure that if you have a problem, you find out about it as quickly as possible, ideally, before your end customers or users call you and tell you about it.

Uh, and best case scenario, you wanna get ahead of them. You wanna be proactive, find out about issues, solve them so you can keep your systems up and running. The the challenge though, Michael, is that, uh, the environment has gotten much more complex, right?

So when you think about data tools, there's just a proliferation of data, proliferation of tools. And, and when I talk to, to customers, they talk about outage of bridges, where they've got all these specialized tools, they bring different teams out to those bridges. Uh, there's a term that we talk about in this industry, right?

Mean time to identification of problems. And, and one customer told me that in their environment, it's actually MTTI stat for meantime to innocence. So all these different stakeholders come to these bridges, and it's all about proving that it's not them, but you ultimately have to get to the root cause, right?

What's actually causing this outage. And so being able to bring all this data together from multiple tools tied together, get to the bottom line problem, and more, more importantly, recommend the action that needs to be taken, that's where technology has changed. That's where AIOps is coming to the rescue.

How proactive can we get? Because sometimes I feel like the application environments are so complex that they're beyond my abilities as a human to keep track of. Or even a team of 10 people still can't figure out what's all connected to what and where all the dependencies are.

And I feel like I need the machines to make decisions on my behalf maybe to, uh, preempt that. But, you know, people get a little uncomfortable with giving up that control. So what's the balance there?

Yeah, I mean, good, good, good point. I always ask customers this question, right? If I'm presenting or in conversations, I ask them about their data volumes.

And the question I ask is, do you expect your data to increase or decrease in the future? And, and every single time without exception, I think there was one time somebody actually answered that the data volume was decreasing. Uh, but they misunderstood the question.

But as long as they understood the question and everyone said, data's increasing, right? So your data's increasing in terms of both volume, complexity, change is increasing. Uh, as teams embrace DevOps and much more agile practices, change becomes a way of life, right?

And that's how you're actually able to innovate. And, and the problem is, those IT teams, they don't get additional budget, right? You're not gonna get additional resources to be able to manage that.

So think of that multiple systems, uh, lots of different types of data. The environment's getting more complex and you're not getting additional resources to scale. So how do you do that, right?

That's really what it comes down to. And that's where technology, AI and AI techniques, uh, we, we collectively call the space AI ops to, to me and from what we talk to customers and what we learn is that it's, it's really existential for it to be able to operate and deal with this complexity. Uh, it's not just a buzzword or a nice step to have.

It is really a set of tools and techniques that help, uh, conquer the complexity, be able to solve those problems without which, especially with digital transformation, IT teams would have no real answer to, to be able to embrace that new world. Initially, it seemed like a lot of folks were skeptical of ai, especially as it applied to IT operations. But is it your sense that, you know, people's minds are coming around full circle here?

'cause they do see it every day in their lives and they see various things that are copilots that are out there. It seems like it's just more common. Yeah, a hundred percent.

A hundred percent. And it's interesting in, in our personal lives, right? There's AI in every aspect of our lives.

Even if you think about just adaptive cruise control, right? I remember driving in the highway, you drive for five or six hours, it's, it's a tiring experience, right? And, and now you're able to, to drive because the cars are doing most of the work, right?

Especially if it's predictable sort of environment and you're monitoring it, right? You don't wanna go to sleep because you have no idea if it's gonna work a hundred percent, but it changes the driving experience, right? In many ways, it's, it's similar when I look at it from an IT perspective, these are tools and techniques that make it easier for you as, as an IT manager or IT leader to, to, to run and solve problems in your environment.

And, and ultimately it really comes down to whether they could solve, uh, specific customer problems, right? Customers don't necessarily care as much how the problem is being solved. They really care about the problem being solved, right?

What is the use case? Can I get to root cause? Can I re-identify and distill through all this noise that I'm getting from all these different tools?

And, and can you point out in real time what's happening rather than me pulling these bridges and, and multiple people trying to figure out what's happening? And that technology's working, right? There's no question about it.

There's been a variety of tech tech technologies that have evolved in the space, uh, especially with, uh, the advent of gen generative ai, gen AI has, has, has really captured a a lot of attention. But it's not just gen ai, right? It's multiple different types of AI techniques that can work together and solve problems around classification, aggregation, summarizing, right?

How do you actually get to the root cause of, of a problem and then summarize it in a human understandable language? And, and that's not just one type of AI typically to get to the root cause of a problem. You're employing graph techniques.

You're looking at knowledge graphs that are doing correlation in terms of time, text topology and relationships of different assets and interface, uh, in, in the environment. Uh, and, and then imagine the power of, of applying generative AI taking that root cause, translating it to something that a level one person can understand. And in some companies, level ones don't even exist, right?

So it's really empowering the changing environment, the changing it structure to be able to make decisions more quickly. Do you think it will become more accessible to a broader number of folks? Because I won't need these huge teams to manage everything.

I mean, I'll still need people, but I can do it at a level of scale that just democratizes a little bit of this in a way that enables more organizations other than the larger enterprises to build a lot of software at scale. Yeah, I mean, that, that is a really good, good question, right? We, we always talk about increased productivity and, and with all these technologies, we, we talk about all the free time that we have.

I, I, I don't think any one of us has free time there, right? I I, I think the reality that happens, Michael, is over time, uh, resources can start focusing on what's value additive, right? The types of things that are mundane, that are, that are time consuming, those types of things.

You have tools that help you do that. And now you're actually focusing on what's important. How does it become more of a competitive differentiation for the business, right?

How do you support agility that becomes the way you operate, right? So I, I think at some level there's probably a, a, a restructuring of who does what, but ultimately it's about making sure your IT employees can work in on what's more, most, uh, differentiating for the business. What is your sense of what comes next for ai?

I mean, you guys have pretty much been working on generative predictive and even causal, and there's lots of different AI models that are out there that have to get stitched together. Um, can we do that in a way that's tied to an automation framework to execute some of these things? Because I feel like the AI only goes so far.

I need something that actually like allows me to put on the rub autopilot. Yeah, that, that's a really good question, Michael. A hundred percent, a hundred percent.

Uh, in fact, with our platform, with the BMC Helix platform, automation is built into it, right? It's, it's, it's all about how do you actually distill through all this noise, get to the root cause, figure out what's happening, making recommendations, right? And making recommendations, figuring out what actions to take, and then taking those actions automatically.

And then automation framework is already built. It's in place that the question is how much to customers trust and turn it on where they can just let the machine do the work. Uh, what we see in practice is it's, it's a process of adoption.

While the technology and techniques are in place, customers need to get to a point where they can see the automation working. They typically track it, they make sure that it works and, and solves their problems. And then at some level, they can turn it on and say, every time you see this type of issue, go ahead and run this automation in response.

Uh, so that's a process that that happens. Typically, the, the, the key thing that's, that's important here, Michael, is just in terms of data, right? Ultimately, when you talk about ai, uh, AI cannot function without data, right?

The data in many ways is the oxygen for ai. Uh, and you have all this data by itself, as we discussed earlier, the data by itself is of no use. You need the AI to make sense out of it.

So the go, the two go together very much in sync, right? And that's how you get value in an organization. Uh, and so ultimately you need to have platforms that can bring in and support huge amounts of data.

Scalability is important. In many cases, you're dealing with problems that you haven't seen in the past. We started talking about monitoring.

In the old monitoring world, it was a very simple problem, right? You are looking for known issues. You typically defined ahead of time.

These are the types of problems I'm going to look for, right? You had certain metrics, you had certain thresholds, and anytime those are violated, you had rules in place, raise an alert, take the action. So life is much simpler in those days, right?

Today, you can't do that. You, you've got much more complex environments, dynamic environments changing with Kubernetes microservices, you've got pods scaling up and scaling down, right? It's constantly evolving as well.

So in those types of worlds, you need to make sure that you've got platforms that can scale, bring in huge amounts of data, different types of data, not just events. You know, when you think about observability, that space has evolved. It's not just about events and metrics, but it's about log data traces and topology relationships between different assets.

And then imagine tying to that service management, data changes and incidents and tickets. Think of the scenario. If you've got a scenario where you understand that there's possibly three different culprits for a particular problem, three different devices, the routers that are po potentially culprits, but you know, that one of them changed over the weekend, guess what?

Which is probably more likely to be the culprit. So think of bringing change into that equation as well. That's what becomes our fault, right?

So as we look at this world and that we live, and it's about bringing new types of data, new ways of processing those, that data, and then solving problems in ways that we didn't think was possible before. So are we moving down the path where the future of an IT team is gonna be, uh, a mixture of agents that I can use to observe and agents that I can use to execute alongside all US human IT folks, and we're gonna, you know, my agents will interact with your agents, or we may interact with different people at different times, but the team itself is just, you know, uh, an amalgamation of AI and humans. Yeah, it's interesting.

I I, I think the lines between AI and, uh, and, and people start to blur that point, right? You get to a point where those are tools, right? Throughout history, we've used technology, we've used tools.

We use tools to be able to do our jobs better. Uh, some of those tools become just part of the, the products that we use may be embedded, and you may not even be aware. In other cases, you are actually very much controlling how it works, right?

And I think you're gonna have the whole range of, of those technologies. Uh, from our perspective, what we look at is how do we make sure we deliver a platform, a platform that's scalable, can bring in different types of data and be able to provide not only recommendations, but also take the actions outta the platform itself. So from a customer perspective, it starts to simplify, right?

They can still have multiple tools in place, but you've got a way to integrate, to be able to orchestrate across those different tools and, and to have a platform that could help solve these problems, uh, in a much better way than before. So when is that crisis moment where people wake up and go, yep, we gotta move forward, we gotta kind of get on this activity because it seems like there's a lot of inertia built into our workflows. There's a tendency just to keep doing things the same old way.

What pushes them forward? I, I, I think that crisis has already happened, right? If, if, if there's companies out there that are ignoring it, then it's, it's at the apparel.

Uh, at this point, these are technologies that are here to stay, right? They will evolve. There's no question about it.

Uh, we think about gen generating ai. We're probably the beginning of, of this whole new, uh, evolution that's happening in many ways, a revolution of, of how we, we use these technologies. And so they will evolve.

And, and my recommendation to companies is to start looking at it. And if you haven't, it's, it's probably, you're probably coming late to the party. Uh, and it really comes down to how do you operate more efficiently, right?

I talked about how environments getting more complex. When you think about businesses being digitally transformed, there's less tolerance for any sort of outage or, or performance issues, right? If you don't address them, guess what?

Your competitors are gonna be out there. They're building those into their systems and your customers are gonna go there, right? So this becomes a very important way of not only dealing with the complexity, but also being able to start training services, start differentiating from your competitors.

Uh, as I look at, uh, you know, how AI is gonna be leveraged, uh, there's always a question, right? Is e is AI going to replace our jobs? What's going to happen?

And, and I remember a quote from Eric Bruno and, uh, he directs the digital economy lab at Stanford. There's a question put to him of, will AI replace lawyers? And his quote was that it's, it's, it's going to be lawyers working with AI that replace those lawyers who don't work with ai.

In many ways, I think it's the same thing in in our IT technology as well. Those that embrace ai, those that embrace products and technologies that have AI built into them are going to be one, the ones that are successful. Uh, it'll be a survival of the fast.

So last question. Do you think that we have limited the amount of software that we were able to deploy simply because the, it is too complex to manage and it's become a self-limiting equation? Yeah, I mean, that, that's a really good point.

And, and, and back to survival. The ones that can solve the problems, right? The ones that can be agile, that can maintain their practices that least capable quickly and efficiently, but also be able to maintain governance, be able to retain and, and ensure that their systems are up and running, those are the companies that are gonna be going to be successful.

And, and that's what we see, right? With the power of ai, it's being able to deal with a world, think of a world where you're putting changes in, and the AI can automatically look at data, understand what's happened. When you applied changes similar to this in the past, did you have outages, right?

What was the impact? What's the risk? Being able to proactively predict the risk of those changes.

And you understand the ones that are deemed pretty straightforward. You just operate and, and let them go through the ones that are more risky. Maybe there's additional due diligence you have to apply.

Imagine that world where human being and AI are working together, being able to deal with this dynamic world where changes are happening and, and the process doesn't stifle innovation. You have the right level of process to work with the innovation and agility that you need to survive. Alright, folks, you heard it here.

Hey, if you don't enjoy doing it, outsource it to the AI folks, right? They're gonna help you and they'll work for you anyway. And you just keep all the fun stuff for yourself.

Hey, Joseph, thanks for being on the show. It's been a pleasure talking with you. All right, and back to you guys and the student.