Insights from AIOps Evolution Weekly | Episode 2
The latest AIOps Evolution Weekly Episode dives into where the term “AIOps” came from, what does it really means, and other philosophical tech ruminations like where AIOps is heading in the future.
Topics on AIOps integration included:
- How AIOps revolutionizes businesses for better productivity and ROI
- AIOps: Past, Present and Future AI Integrations
Let’s take a deeper dive into the article topics and the takeaways from Bill and Sean.
Topic 1: AIOps revolutionizes businesses for better productivity and ROI
By: Shamus McGillicuddy, Research Director for the network management practice at Enterprise Management Associates
Enterprise Management Associates (EMA) published “Revolutionizing Network Management with AIOps”. The report found that 90% of the surveyed networking pros (309) believe applying AIOps to network management can lead to better business outcomes for an overall enterprise. In short, AIOps-driven network management can help a business run better. It translates into better employee productivity, improved customer experience, and revenue generation. But how exactly does AIOps enable this boost to the business?
Network management benefits
The EMA survey respondents identified the top benefits of AIOps-driven network management in the following order:
- Network optimization
- Operational efficiency
- Improved security/compliance
- Network resilience
- Cost reduction
Optimizing Existing Toolsets
According to the survey, about 91% of respondents are hoping that AIOps can address problems with their tools, including preventing conflicting or inaccurate data, prioritizing real-time insights, and eliminating tool fragmentation.
Gentle reminder: Don’t expect AIOps to fix all “tool problems”
Networking pros should not expect AIOps to be the “magic toolbox” that fixes all problems. Most respondents who had successful AIOps implementations reported that their AIOps interest was not driven by network management tool problems.
This suggests that successful users of AIOps are focused on transforming network engineering and operations, rather than addressing challenges with their network management tools.
Sean and Bill’s Take
- Sean and Bill both noted that the EMA study makes it sound like 90% of enterprises are using AIOps; it’s more like the vast majority (90%) of companies are only investigating AIOps as a potential solution, in their experience
- The reality is that roughly 10-15% of Windward clients are actively implementing AIOps
- The article touches on the “tech tale as old as time”: If a vendor had functionality where we could marry their solution with other tools and build AI around that, it would be useful. Vendors should look for ways to consolidate panes of glass with other tools, either through acquisition or ingesting and sharing data.
- Vendors have changed. The use cases have changed, but the argument is still the same. And you could make arguments on either side that, well, are these large vendors who have to go wide in order to continue to drive the massive amount of revenues that they have. Do they start to innovate through acquisition or do they actually continue to build and innovate internally on these technologies?
- CONCLUSION: In the next few years, the AIOps market may double or triple with more products and features touted from side companies, but those will most likely be absorbed into larger enterprises like ServiceNow or IBM.
Topic 2: AIOps – How we got here and where we’re heading with AI integrations
By: James Maguire, Editor-in-Chief at eWeek
The irony of our technological advancement as humans is that we have created systems that are so complex, especially in the IT domain, that we are overwhelmed. With cloud computing ever-expanding, data analytics superseded by real-time streaming data, hackers galore, and budget constraints, IT managers and department heads are crying out for help! The answer has come in the form of artificial intelligence (AI) for IT operations – or AIOps. But how did we get to AIOps, how do we use them and what is in store for us and AIOps in the future?
What is AIOps: Gartner Coins an (Awkward) Term
Although humans created IT systems that have overwhelmed us, we have also been advancing in creating a secondary system to help manage the original IT system. This secondary system has many names (AI for IT Operations, MLOps, etc), but the one that stuck, AIOps, was originally coined by Gartner in 2017.
The Uses of AIOps
The challenge of many IT systems is not having enough oversight or foresight to make predictive analyses and decisions on how to protect the current IT infrastructure.
AIOps fills in those gaps in knowledge; it provides predictive, problem-solving approaches to enable proactive reaction.
Top use cases among companies include the following:
- Predictive Alerting
- Root Cause Analysis
- Prioritizing Events
- Predictive Outages
- Service Desk Ticketing
The Future of AIOps – Two Shifts
Although AIOps is in a nascent stage of adoption, experts and analysts are predicting exponential growth in this market sector within the next 5-10 years.
- Exponential increase in AIOps
- AIOps will leverage more and more AI going forward, exponentially expanding the support it offers to IT systems
- Companies that don’t deploy AIOps may not be able to compete
- The AIOps market, hovering around $15 billion in 2021, is expected to hit $40 billion by 2026.
- The meaning of AIOps will change
- As it becomes more commonplace, the term AI will drop from the AIOps; it will simply be part of “operations”
- Companies will choose AIOps systems that will scale on multiple levels in the long-term
Sean and Bill’s Take
- One highlight of the article is some of the philosophy behind AIOps – Simple tasks done manually can really add up as systems become more complex. AIOps is the strategy that conglomerates all of that and simplifies these massive tasks and events.
- The use cases for AIOps are exactly the same as they have been for the last 20 years, but the infrastructure has gotten much more complex.
- While we focus on AI for IT Operations (AIOps), it will converge with other systems and departments in enterprises. AIOps / IT Operations personnel should be looking into this for the rest of the CIO organization.
- For AIOps to work well, we have got to really define and understand the problem. Then you can decide what’s the best use of technology to solve it.
Catch the full details from AIOps Evolution Weekly
Transcribed from Temi without edits
Speaker 1 (00:01):
Welcome to the AI ops evolution weekly broadcast. This series features discourse on topics pertinent to the C level conversation revolving the integration of artificial intelligence into the ever evolving it operations landscape, discover trends, and actionable tips for digital transformation, as well as the latest news on how AI ops is transforming business and government operations to create flow in their organization. And don’t stay in the dark on the latest industry news. Join the AA ops evolution community to stay connected and subscribe to the firstname.lastname@example.org. Now let’s join our hosts, Sean McDermott and bill Driscoll. It thought leaders, visionaries, and the voices on the other end of the mic for the latest episode of AI ops evolution weekly.
Speaker 2 (00:53):
Welcome everybody to the AI ops evolution weekly broadcast. My name is Sean McDermott and my co-host bill. Bill. How you doing on this Friday? I’m good.
Speaker 3 (01:04):
Good. I’m looking forward to another week in a road going up to New York. So just visit
Speaker 2 (01:08):
Family. That sounds fun
Speaker 3 (01:12):
Out of town. Okay, go ahead. Cooler weather.
Speaker 2 (01:17):
Yeah. So do you want to, you want to talk a little AI ops stuff before the weekend? Nothing awesome. Okay. All right. So we got a couple of articles here, a couple of topic we want to talk about. So let’s start off today with so a network world article came out really referring to an EMA study that basically said that EMA study found networking pros who reported the most success with AI ops pointed to improve security compliance as a potential benefit. So this is a pretty interesting article. I think I say that every time, this is an interesting article, I guess we wouldn’t be talking about it if I didn’t find it interesting, but I think one of the things that was interesting to me about it is, and we’ll get into a little bit more, but just on the surface and report found that 90% of the surveyed networking pros believe applying AI ops to network management can lead to better business outcomes and overall for an overall enterprise. I found that interesting because that makes it sound like 90% of people are doing AI ops. And I don’t think that’s true. The I think the vast majority of companies right now are, at least in my experience are investigating AI ops as a potential solution for them. And so is this, is this a hope, you know like 90% of people hope that this is what’s going to turn out with AI ops?
Speaker 3 (02:45):
Yeah, I mean, I think so. I think, I think I do agree 90% are at least looking at AI ops, they’re evaluating it. We’re talking to customers every week just about doing an assessment or doing some experimental things. And so, yeah, I’d say I agree with that 90%, I think 90% are not, definitely not. In the middle of the implementation of any kind of AI ops that might be that might be in the mid, I don’t, we don’t have numbers, but you know, that could be 40, 50% I would say.
Speaker 2 (03:15):
Yeah. Yeah. I, I think that, you know, in our clients, you know, we’re looking at probably 10 or 15% are actually actively in implementation of AI ops solutions and mostly AI ops monitoring platforms. Do you agree with that?
Speaker 3 (03:33):
Yeah, I think you know, we’ll sort of see, I think we’re going to talk a little bit later about the vendor landscape, but you know, I think some companies that already have a vendor that’s, that’s invested in, you know innovating or, or investing in sort of AI ops capabilities. I think they’re definitely working with those vendors and maybe they’re doing it in a, in a POV status, or maybe they’re rolling it out to some extent, but, you know, I think the range of what we define as AI ops is pretty broad and, you know, I think it boils down to a handful of things. And I think some vendors are doing some point pieces of it and other vendors are sort of trying to go all in and really be nothing but an AI ops platform. And I think, I think it’s probably a low percent that are, that have invested in that, you know, pure play, you know, a hundred percent AI ops platform, it sort of bolts in and, and, and, and sort of fits with the rest of your ecosystem. I think it’s more individual features that are a, you know, machine learning, like, or even maybe some machine learning that, that I think a lot of customers are looking at implementing right now.
Speaker 2 (04:37):
Okay. Yeah. Another part of this article, and we can have a whole discussion on this and maybe we will in the future they quoted one person, one of the surveys, we have so many, so many tools and so many gooeys and every single tool is doing just one thing. If a vendor had functionality where we could marry their solution with other tools and build AI around that it would be useful. Vendors should look for ways to consolidate panes of glass with other tools, either through acquisition or ingesting and sharing data. Yeah. So I’ve been doing this a long time, right? I started our, my company twenty-five years ago and I was in operations before that this is like literally the age old as far as I’ve been, you know, almost last 30 years is best of breed versus platform place. Right. And we’ve seen this story before of, you know, do you go all in with CA do you go all in with BMC?
Speaker 2 (05:35):
Do you go all the way with HP back in the day to now it’s like, do I go all in, on service now? Do I go all in on Splunk? And so it’s like the vendors have changed. The use cases have changed, but the argument is still the same. And you could make arguments on either side that, well, are these large vendors who have to go wide in order to continue to drive the massive amount of revenues that they have do they start to innovate through acquisition and, and or do they actually continue to build and innovate internally on these technologies? In my experience over the last 20 years, a lot of these large vendors, they they innovate through acquisition, including a company that I sold to BMC and eventually you know, these tools start they’re written in different technologies, they’re architected differently. And they start kind of trying to glue them together into a platform, play this kind of best this best, or this full blown plat platform. But when you lift up the covers, it’s still somewhat of an architectural mess underneath. So what are your, what are your thoughts on that? And I guess, I think we should have a whole discussion on this in the coming weeks. Yeah, exactly. Yeah.
Speaker 3 (06:54):
I think, I think it’s sort of, that’s a great segue to the next topic, which is sort of the marketplace, but before I sort of go there yeah, there was one last point I thought was interesting, that was written on that, you know, that, that Seamus Magilla Katia network world pointed out which was a lot of, a lot of organizations are looking at the tools that they have. And like you said, all these disparate tools and looking at what are the gaps that we have in our tools, what’s the inaccurate data, or, you know, w what are we not getting in the tools? And, and kind of focus focusing from that point. I think a lot of vendors that will talk to that, that we’ll refer to are really looking at where’s the puck going. So in hockey, you skate to where the puck’s going.
Speaker 3 (07:36):
And I think what you’d really need to look at what he recommends is you got to look at your engineering, you gotta look at your operation, you’ve gotta look at the technology that’s coming and the infrastructure that’s part of that, your software defined network, function, virtualization, and containers, and the cloud, and say, what, what does that infrastructure going to look like? What are those services going to look like, and what are our needs, and then start to build towards those needs. Instead of just looking at your, you know, backwards, almost at the tools you have and what gaps you think you have, because fill in those gaps may not get you where you’re going. And so, you know, that that sort of leads to this, this where’s the marketplace going. And I think I’m not going to name any, any vendors because they’re, they’re really emerging, you know, coming out, but I’ve read articles just in the last couple of weeks from several that are really they’re really press releases, but they’re, they’re, they’re providing some insights that are really trying to find gaps that the large vendors have and figuring out how do we play, right?
Speaker 3 (08:34):
And, and all of their, all of their messaging is like, you don’t need to, you don’t need to replace anything. You just bring us in as an overlay, feed all this data. And then we’re going to give you the insights where you have patented, you know, unsupervised machine learning, or we have a, that you know, predictive algorithms or, or patented detections. And by bringing all of this data back into our system, we’re going to be able to kind of provide those insights, or we’re going to be able to provide that sort of real-time identification of where you really need to focus. And, you know, I, I think a lot of these vendors are going to be struggling with growth, you know, going into a large enterprise and the volume of the data, the disparity of the data, you know, I think some of them will find are probably one trick ponies or maybe three trick ponies. But you know, it’s going to shake out and they’re all sort of contributing to this growing market that really could double or triple in the next few years. And some of their innovations are going to die out and some of them are going to be acquired by the service now’s in the Splunks and the IBM’s and whatnot.
Speaker 2 (09:39):
Yeah. I think, you know, having, having gone through this, right. You, what you’re going to find is that a lot of these products that are emerging really aren’t products, they’re more features, and those features will ultimately be acquired. Those, those companies will be acquired for those particular features. And I think that’s a very common thing. The other thing I, I think that we need to keep in, in, in check here is, you know, the, the companies that are looking at AI ops, so companies that are always really on the bleeding edge of adoption are usually the largest enterprises that have the largest issues around scale. And w we’re seeing it right now with AI ops and, and these large, large enterprises that are just creating massive amounts of data and have to start employing AI technology to process it. And to, for all the use cases that I think we’re going to talk about in a little bit the but you know, a lot of these products, you know, when it come to market, I don’t know if they’re exact and in my experience geared, you know, ready for the enterprise.
Speaker 2 (10:50):
Right. And they’re kind of ready for mid-level enterprises. And then they go into these large enterprises and they kind of just get clobbered. So we’ll, we’ll see how that shakes out. I think you’re right. There’s this a lot of vendors coming out. I’ll do a little bit of a shameless plug right now, the AI ops evolution podcast, which we released season one it’s out, we’re actually in production right now of season two. And in season two, we’re actually focusing very much on the vendors and we’re bringing in every single vendor that we, you know, kind of the top 12 or 14 vendors, we’re still formulating the number and we’re interviewing each one of these vendors and giving them a chance to kind of talk about their vision, what they’re doing, what their product, what they see is where AI ops is going, where they’re meeting AI ops, what their differentiators are. So I think that’s going to be a really useful podcast for people that are thinking about AI ops and trying to sort through some of this, this vendor technology, but for this, for this broadcast, we don’t really want to kind of go too much vendors and give plugs and things like that. But any other final thoughts on that?
Speaker 3 (11:58):
Yeah. The one final thought is I think you brought up the point that these, a lot of these newer products are features. I think they’re contributing, they’re disrupting, right. But you know, they’ll name big time. You know, companies, we’ve all heard of fortune 100 companies that are their customers, but when you drill in, they really found a specific use case. They’re all trying to wedge themselves in and say for this new business unit, this new 5g or edge, or or containers or Docker, they’re, they’re really just laser focused on that, trying to get in and wedge themselves in and solve that use case. And they hope, you know, they’ll continue to innovate and grow. And I, and I do think from a company standpoint and enterprise standpoint, that is the place you want to start to innovate and experiment and try out some of these vendors that are bringing in some, some things that you’re not going to see from the big vendors for probably a few years out.
Speaker 2 (12:48):
Yeah. Yeah. It’s, I mean, the one thing I tell my clients right, is, you know, we say this all the time, AI ops is a strategy, right? It’s not a technology, it’s not a platform and you have to look at it over a multi-year. So you have to have a vision of what you’re trying to do. You have to have a starting point, you have to think about starting small and, and how you’re going to build momentum through AI ops use cases and kind of knocking them off one by one while people get involved, because there’s so much there’s a lot of, there’s a lot of myths around AI ops, too, right. That need to get dispelled through that process. And, and we’re just in the beginning, you know, I think that that’s there and that actually is a good lead in to our final topic.
Speaker 2 (13:31):
And Roy, the AI ops the history uses and future of AI ops. So article written by James McGuire of E week. So this, this, this one kind of caught my eye. It it’s an interesting article in that it goes a little bit into the history and a little bit into the future. And some of the current use cases right now, I actually wholeheartedly agree with pretty much everything in this article. And they talk a little bit about AI ops and where the term came from. It was actually coined from Gartner, which is interesting because we work with a number of analysts and they they’re all kind of trying to take their own spin on AI ops because they don’t want to call it AI ops to give, you know, a competitor like Gartner, you know, kudos. But in the end, AI ops is, was coined by Gartner.
Speaker 2 (14:19):
And that’s what everybody’s calling it. And it gets a little confusing when other analysts use other terms, you know, like, well, what does that mean? And they’re like, you mean, say I ops I’m like, oh, okay, great. Talk a little bit about some of the use cases. One of ’em the chief product officer from BMC talked about some of the use cases, the outlined five use cases. I think those were dead on, we’ll talk about those in a second. And then essentially, you know, what are the two shifts of AI ops coming into in the, in, in the future? So what are your thoughts on, I know you read this article, what are your thoughts on it? And I, you know, initially, and I think we talked ad nauseum about why AI ops is even a thing, right. And, you know, the processing of data, data streams, and massive amounts of data. I don’t think we need to get into that too much, but I think it’s interesting that, you know, kind of talking about the use cases that I’m the chief product officer for BMC brought up you know, what are your thoughts on that? Why don’t you outline these use cases and give me your thoughts on that.
Speaker 3 (15:25):
Okay. I’ll actually, you know what I didn’t take, I don’t have the article in front of me just to look at what those use cases are.
Speaker 2 (15:31):
All right then, and then I will take
Speaker 3 (15:33):
In front of you. Why don’t you do that? I
Speaker 2 (15:35):
Had some thoughts
Speaker 3 (15:36):
In front of me there. But I, I agree with you before you do that. I agree with you. It was a very well-written article. And I think you know, I like the combination of pragmatic pragmatism, you know, where things are going and kind of real use cases, but he also gets pretty philosophical about sort of, you know, the nature and where technology is going as a whole in complexity. You know, and, and really simple things that are getting solved. You know, they, they seem simple on the surface, but you compound them and it creates just massive amounts of complexity. So I’ll, I’ll leave it at that. Like I said, we’ve talked about kind of the why of AI ops, but I liked the way he explained some of that. But to some of the use cases you want to go through those.
Speaker 2 (16:17):
Yeah. Yeah. So these are use cases that you, you know, you work on on a daily basis. So you know, and, and again, you know, as we talk about the beginning of AI ops and why we were even talking about it, it’s about processing mass amounts of data. So when you get into the operations world, what are the areas that process massive amounts of data create law and it’s monitoring, right? Monitoring is really the, kind of the primary area that just creates tons of data, whether it’s monitoring infrastructure, monitoring, applications, monitoring containers, and microservices and things like that. So the five use cases you talked about were predictive alerting root cause analysis, prioritization of events, predictive outages, and service desk ticketing, which, you know, it’s kind of funny, right? Cause again, you know, we’ve been around the block for awhile and I’ve been doing this for 30 years and I can remember literally having conversations in the early two thousands, you know, with people and working with them on tools like micro muse, and an HP OpenView saying, Hey, you know, this is all about root cause analysis, right? And this is about prioritizing events and, and, and using this data from micro views to do predictive analysis. I mean, that was literally the story 20 years ago. Right. same exact use cases, use cases haven’t changed, but now we’re applying AI to it.
Speaker 3 (17:45):
Right? Yeah. I think it’s interesting, you know, I did HP OpenView back 20 years ago and then, and then IBM’s typology that came out of micro. And before that Riverside I’ve implemented those at some large telecoms banks medical pharmaceutical or medical facility. You know, and I think, you know, the big thing that HP did was downstream suppression, right? And it was like, we figured out downstream suppression suppression, when all those pings, you don’t have to, you know, paying failures, you don’t have to worry about, cause we figured out where it is, we’ve solved the root cause analysis. And it’s like, not really, there’s the, you know, networks are being configured at the layer more complex and then river soft and which was Mike Hermes. And then IBM came along and they figured out how to build a model that looked at connectivity and relationships and more the containers.
Speaker 3 (18:33):
It isn’t just a router, it’s a router of modules or it’s V lands and there’s a whole hierarchy in there. And they figured out how to get suppression that way and became all right, we’ve solved the root cause analysis. And now in today’s world, you’ve got multi cloud and containers and virtual machines that go up and down. And so, so the world of problems to solve is become so much bigger in today’s today’s you know, method is, let’s just enrich it with lots of information and we’ll look at the textual content of alerts and look for patterns there. And so, and I think this is just one step in that evolution, right? As far as all the different algorithms in machine learning that are going to start to be applied and kind of be brought to the forefront of, of how you solve these problems. But yeah, there’s five use cases you just mentioned. I didn’t note them down cause they’re really the same five use cases that we’ve been seeing for 20 years. Right. I don’t think that, I think that’s kind of a known, right.
Speaker 2 (19:32):
Yeah. Kind of a funny, funny story is that when so we founded, I founded a company called real ops back in 2004 and it was the first runbook automation platform on the market. And I remember when we were raising money and talking to investors and a lot of the investors one particular group who actually ended up who actually ended up investing in us basically said, Hey, I was doing this a long time ago. Like they haven’t solved this problem. Like that was literally like 17 years ago. So yeah, the use cases are exactly the same. It’s just that the infrastructure is getting so much more complex. Right. And w I kind of laugh about the use cases not changing, but in the end, that’s really what operations is all about. It’s it’s except that the infrastructure that we’re managing 15 years ago when you had, you know you know, switches connected to routers connected to, you know, in servers connected to a, a switch, it was more straightforward.
Speaker 2 (20:40):
Now you’ve got like you said, cloud services, you’ve got virtualization, you’ve got instances coming up and coming down. So your topology is almost changing instantaneously. And you’ve got all these new layers of, of, of software components that are all kicking off information. It’s just harder and harder. So last thing before we kind of head out for the weekend the last part of the article, which I thought was really about the future, right? Two shifts and in AI ops, one about AI will exponentially increase, which we kind of mentioned before. We’re just in the beginning throes of this. And and I think that, you know, three years from now, we’re going to see very different things going on AI ops. And the other thing is AI ops meaning is going to change. I think where I stand on this, this thing is I think that the concept of AI ops, if you’re, if you’re thinking about it as a monitoring platform, yes.
Speaker 2 (21:40):
AI ops is going to change. The meaning is going to change. When you think about AI ops as a strategy as we do, then I don’t know if the meaning changes. You just accept the fact that AI is going to start embedding into all kinds of different parts of your infrastructure and your, and your, your management systems, whether it be virtual chat bots, like service now is working on now, or, and we talked about last week, embedded AI into salt, into hardware platforms, into the bedded, into the bedded, right into the right into chips of, of, of, of the servers and things like that. I think that’s going to, that’s going to change, which basically goes back to shift one that it’s going to accelerate, you know, very, very fast.
Speaker 3 (22:32):
Right. All right. Yeah. I’ll make a couple comments on what he said about the future is a quote on me. He said he sort of used AI for it, operations, which is what IOP stands for now to just becoming more broadly AI operations and becoming, and I’ll quote a single all encompassing AI system that doesn’t just help monitor it operations, but also forecast business spending employee retention rates analyze the success of marketing campaign. So he really makes a pretty bold step and says this AI, that we’re kind of focusing on it operations, as well as the AI and all these other problem areas are all gonna converge. And I think we, we kind of touched on that last week and that I had said, you know, that the AI I ops or it operations people should be looking into the rest of the CIO organization.
Speaker 3 (23:19):
The other thing he mentioned is just the convergence of AI ops and robotic process automation. You know, and I think my big takeaway there was really, I think people have to get really good and our customers and we as well at defining problems very well. And I think as you define the pro those problems, you’re trying to solve very well. I think then you start to educate yourself on machine learning and AI, and what are the what’s the level or what’s algorithm, or what’s the approach to solve the problems? Because I think it isn’t, you don’t just apply AI, AI, or robotic process automation in a blind way to the problems. You really got to understand the problem that you’re facing and what’s the best use of technology to solve it.
Speaker 2 (24:02):
Well, I think that, I think that wraps up this week and I appreciate your time. I hope you have a great weekend. I hope everybody watching has a great weekend and get some time off. And we appreciate you guys listening in. I want to thank our sponsor Winward consultant group. The number one global leader in AI ops for sponsoring this, this broadcast. And we will see you next week. Thanks bill. Have a great weekend. All right.
Speaker 3 (24:30):
See you, Sean. See you next week. Bye.
Speaker 1 (24:33):
Thank you for joining us in this conversation about AI ops and the future of it. Operations interested in a deeper dive on these topics. Check out our resources email@example.com and remember to subscribe, to get the latest episode.