Improve Efficiency, Reduce Risk: Latest BMC AIOps Innovations and Future Vision | BMC Connect 2024

Hello everybody, I'm Mike Ard and we're back at six five on the road in Las Vegas at the BMC Connect event. And we're here with my friend Kiron, and we're gonna be talking about AI ops because, well, it is all the range these days. Karen, welcome to the show.

Thank You so much, Mike. Hey, We've been talking about AIOps for some time now, and I guess my question is, um, it's been a journey. Sometimes people, you know, got excited about it, but then there were hits and misses.

Where are we right now in the maturity of the concept? Uh, so I think we are definitely at a point, Mike, where it's starting to get real, it's starting to get real where customers are seeing value out of it, real business outcomes. Uh, some of our customers last week in Stockholm, uh, talked about they reducing their incidents by almost 85%, uh, compared to just six months back.

We are seeing customers who are able to reduce the time it takes them to respond to an incident significantly in the morning. I'm not sure if you saw Ali talk about it's now taking from months, just too few minutes to be able to get changes done in the environment. All of that is AIOps and I think I said it's exciting times.

There's still some, some, some way to go, uh, as we mature it and completely industrialize it, but I do feel we are doing a great job there and BMCs up top there, uh, in that area, There's an old AI joke about it's one thing to be wrong, it's another thing to be wrong at scale. You guys have brought out some new tools that kind of help me figure out what the risk is for change. So walk us through that.

Absolutely. So, uh, this is, uh, what's the problem that we're trying to solve here? Uh, with DevOps and a lot of automation going in there, uh, customers organizations are, are making changes to your production environments, sometimes hundreds of times a day.

Uh, it's a, it's a long laborious task, but at the same time it's very risky as well because, uh, if you see more than two thirds of of outages that you typically see in, in customer environments are because of an incorrect change that was triggered whether scheduled or unscheduled. So what does change risk advisor within our AIOps tool do today? AIOps is getting data, not just about metrics, about events, about services, but it's also getting information about incidents, uh, all the tickets that are out there.

So we're doing two things with our change risk advisor. Number one, looking at the service management data. So what kind of changes were similar to this change that has been submitted?

That's number one. What was the outcome of some of these changes? Did these changes fail?

Did they cause any downstream effect for the services that they're managing? So that is one element of it. Now, what is AIOps bringing to table is a dynamic service model.

So we are able to map the cis, both the hardware assets as well as the software cis into a business service model and give a view to the user of what is the context. So your server is connected to a network device. The network device is connected to a edge router and so on and so forth.

But what we're able to do is once you get a change request, we are able to map the change for the CI onto the service model, and then you are able to determine what would be the business impact of that change. So not only are you looking at the past data, historical data, but now you are looking at the potential realtime impact to a service. So we have had scenarios in our pilot customers where three changes are affecting the same service at the same time, we were able to uncover that as part of our beta testing.

Now what does this mean? Those are risky changes. We would want to take some time, assess those, maybe get them through a cab and then approve them.

There might be simpler changes which are, which are just not as risky. And then you could just automate that and improve the speed of how you, how how changes are made to the environment. So that's, that's what we're trying to get to, both the historical element as well as the context of the business service there.

Mike, If I happen to take out a service, will you tell me how much it will cost the business? I mean, is there like a dollar or something that goes with that or? Uh, So that is, that is, that's, that's what we are trying to now get to mapping service ops, which is what is, as an example, what is the SLO associated with a particular service?

Now, if you, if there's an impact to the service today with AIOps, we are able to tell the technical impact to that service. With this service ops concept and AIOps bringing services together, we are able to say, oh, this service is a top service, your checkout service, if this gets impacted, you are gonna lose probably $5 million uh, a day if you're not able to fix that in time. So we are trying to bring the revenue element to this as well.

So eventually service ops and finops will all come together under AOP Ops? Yes, absolutely. So there's an element that is not in the current release, but we are working on from a roadmap perspective, a cost optimization.

So today we are able to pinpoint that a particular infrastructure could run into a capacity problem in some time. So we're able to do that now. We are trying to take it to the next level.

Would we be able to recommend to the users, okay, this is the kind of CPU scale you should use for this app application. This is the memory size you should use. This is which cluster should that application run on?

So we're trying to get to more level of specificity that would enable us to do a lot of cost optimization, finops, as you call it, uh, as well Mike. So again, uh, the, the gamut of use cases that now we are getting to because of the data being at one place, uh, it's phenomenal. It's amazing.

I'm very excited. I've been doing this for 25 years, but I've never been more excited about this before. I feel like we've turned, made a turn here.

Last year was all like fear and loathing of AI and I'm concerned about my job this year. I talked to folks and they're like, I wouldn't wanna do this job without ai. It's too tedious.

It burns me out. And now they're looking at it and saying, you know, these AI agents might be my best friend. So have we kind of come full circle?

Absolutely, yes. And, and again, it's not just as I said, the agents on ground, but it's the managers, the, the CIOs as well where we are seeing the change in stance as well where there was an element of worry or does this mean my team size reduces? Does it mean I need to let go of people?

And what we have seen, and you might have seen some of the, the customer success stories as well, where yes, absolutely yes, you have been able to reduce the amount of effort, not necessarily the team size. We reduce the amount of effort going into solving those problems and now that effort, those people are contributing to innovation. So yeah, making sure that you're getting new products, new services out to the market.

So it's absolutely, absolutely true, Mike. Yeah. Now one of the most stressful areas of any ops is security.

And you guys are now talking about vulnerability ops or ops? Ops. Yeah, I mean I think not maybe we can see it for the first time.

One ops. Yeah. So, uh, so one, one clarification.

I think I just wanted to call it out. And this is something we discuss internally as well. Uh, our intent is not to be the top security tool in the, in the, in the marketplace with this capability.

So what we're trying to do is we have operations manager as a persona and what is our operations manager trying to do? The operations manager is trying to get all the relevant data that could impact the IT assets, the business services that, that they are responsible for. Now, as we are looking at all the data vulnerabilities, some of the security information is a very, very natural extension to getting that data.

So what we're trying to do, we are trying to get that data from Tenable Rapid Tools that have this data already. So we are integrating with those, those tools, getting the data. Now what is the value that we are providing with this solution ops, because of the service model that I talked about earlier.

Now we are able to take the vulnerability and map, okay, vulnerability number one has impact on checkout service. Vulnerability number two has an impact on payroll service. So which service is more important?

So we will be able to assist our security colleagues working closely with our operation colleagues, the SecOps will where we are able to prioritize some of those, those vulnerability because the vulnerabilities are in hundreds, sometimes thousands as well. That's first element. Second, using our Helix GPT, we are able to provide recommendations on how to fix some of those vulnerabilities as well, including generating code.

So we are able to generate Ansible or pull code that the users can just automatically run and get some updates done as well. All right folks. You heard in here AI ops, it's all about reducing everybody's stress.

You've been watching six five on the road in Las Vegas. Kiran, thanks for being on the show. Thank you so much Mike.

Alright, We have a ton of other episodes to check out. So by all means, go through that list 'cause you're gonna have more great conversations like this. We'll be back in a minute.