?

TEC37 E15: Leveraging Cisco Products to build an AIOps Ecosystem

39:21
46
Plays

Join host, Robb Boyd and WWT’s Tanner Bechtel, Chris Weis and Neil Anderson as they discuss how WWT has developed a true AIOps reference architecture, and a strategy that leverages a multi-product and product agnostic approach to an AIOps Ecosystem. They will highlight pieces of the AIOps infrastructure including AppDynamics, ThousandEyes and Cisco Intersight that integrate to create a collaborative ecosystem. This episode is sponsored by AppDynamics.

Please view transcript below:

Robb Boyd:                   Welcome. Today's show takes us into the world of AIOps or artificial intelligence for operations, as we take a look into the reference architecture that World Wide Technology has been developing. We have guest experts here to highlight core elements like AppDynamics ThousandEyes, Cisco Intersight, and a few more, all integrated to create a collaborative ecosystem. Welcome to TEC37, the podcast, covering technology, education, and collaboration from World Wide Technology. My name is Robb Boyd.

                                    Welcome to TEC37, gentlemen. Nice to see we've got another cross section, a nice cross section, I would say, of expertise here, Chris Weis. I'm not sure what your title is. I'll get back to you in a second. Neil Anderson, Senior Director of Network Solutions, and Tanner Bechtel, Global Director of AIOps and APM, I assume that's application performance monitoring.

Tanner Bechtel:            It currently is, my friend. Yeah.

Robb Boyd:                   Okay, good, good, good. All right. You're all from World Wide Technology and you've got some different backgrounds which will all come into play here, but first let's start with Chris.

                                    Chris, forgive me, but what is it that you do? I know you something in compute, but I've heard you're a master of quite a few things.

Chris Weis:                   That's a great question. Yeah. So my name is Chris Weis. I'm a practice manager on our global engineering team and I lead up our software-defined infrastructure practice.

Robb Boyd:                   Okay. Perfect, perfect. And then, I don't know if I did justice. Neil, you've been on the show before, but I don't know if ... You can give a 20 seconds on what you're responsible for there as global director on the networking side.

Neil Anderson:              Yeah, absolutely. We're responsible for all the networking solutions. So pretty much anything that moves a packet for including wireless, data center, networking, campus switching, et cetera. That's our practice.

Robb Boyd:                   Okay. Perfect, perfect. And Tanner, you oversee everything AIOps, I believe, and where that actually blends into application performance monitoring. Is that a fair way to put it?

Tanner Bechtel:            Certainly. Yeah. We focus on the app-centric AI model for IT operations.

Robb Boyd:                   Perfect. And obviously, so Tanner, we had talked before, but this is a lot of what we're talking about here. And we had a few other shows on this where we walked through a couple of versions, but World Wide Technology has taken a very concerted interest in physically building out and testing with the myriad of vendors that all claim some part of an AIOps, but also, due to no fault of their own, it's just the way the market works, but it can be confusing. And, so in the same way that we've done this a number of times, I thought it might be good to kind of look at the definition of AIOps, but I'm kind of curious to get it from each of your perspectives. I'm going to go back to Chris. Chris, how would you describe AIOps?

Chris Weis:                   Yeah, so, I primarily live in the infrastructure space, so when we talk about AIOps in my world, I'm really thinking about ops as it relates to infrastructure itself. So, on the infrastructure side, we typically worry about things like performance and performance optimization. So, we want to make sure that workloads are running optimally, wherever they're running, but we want to make sure that we're not over-consuming things or violating some sort of a policy or consumption policy inside of our organization.

                                    So, we leverage today, it's a lot of manual intervention from humans to make sure that goes the right way, but AIOps gives us the ability to look at a larger data set and make sure that we're making the right business decisions and keeping things running the way they should be across the board.

Robb Boyd:                   All right, good. So I hear infrastructure and I hear a certain level of focus and the things that you're primarily focused on, but I don't think that's, of course, all there is to it. And not that you were claiming that it was, but Neil, how do you go about describing and feel free to piggyback on anything on all the elements that are still part of how you would come about answering that question.

Neil Anderson:              Yeah, I would say network connectivity is getting more and more complex. There's multiple paths to get to the cloud where my applications are living or in my private data centers. So, applications have moved, users have moved, and the network is still trying to connect to all those things. And so, what we have is we have a lot of different data sources and visibility trying to provide insight into where might that connectivity be going wrong?

                                    And I look at AIOps as a way to bring that together, to make informed decisions about where really is the problem. Is it a connectivity problem, a data center/infrastructure problem, an app workload problem? Helping to identify what's the real source of those problems, that's kind of what I think of when I think of AIOps

Robb Boyd:                   Tanner, do you have anything that contrasts with anything being said here, or how would you bring it all together?

Tanner Bechtel:            One of the great things about AIOps, if it's consumed in the right way or if it's planned the right way objectively, is that it does consume all of these [inaudible 00:04:40]. And when we first started our practice and focused on ways that we solve application problems, we looked at it from a performance point of view. APM, how do we watch the application and fix the application? But when you take into account the entire data center, the entire infrastructure enterprise, you see three things. You see application performance, monitoring and management, you see network performance monitoring and management, and you see application resource management. And so, all three of those things are real, they're all wrapped in a security layer, but at the end of the day, AIOps is not a single definition. It is. People always ask me, "What is the simple definition?" We are trying to isolate high value decision-making.

                                    So, if I'm someone that runs part of the data center, or I own an application, or I own a network, I could be in Neil's shoes inside of one of our customers. I don't care what's on the highway, whether that's an app or that's a commercial packet from somewhere else. It doesn't matter. My job is to maintain that stretch of road or that piece of my network. And so, if we look at AIOps as the model by which we are going to start making intelligent correlated decisions about how to deliver resources to our customer, whether they're booking an airplane ticket, or they're trying to check their gift card balance or whatever they're doing, they're touching all of these things. And so, when failure occurs, it doesn't always happen at the code level. It's not always contained inside the app.

                                    So, AIOps is an extension of all those pieces across all of these regions. And we just got a fraction. You know what I mean? We've got Chris and Neil are both experts in their field, and they're a component of that, but AIOps really has the growth potential, if executed the right way, to consume the entire data center. If we can isolate problems and events and failures and metrics, and get smarter and execute with more intelligence across the entire enterprise, that is the true big picture definition of what AIOps mean.

Robb Boyd:                   Well, it feels like we've, as an industry, have come into AIOps, I don't want to say, "Accidentally," but we've kind of fallen into it as we search for answers, because all of a sudden, I think so many of us feel like we're losing control. Our applications are in so many different places. The data center is not in one fricking location like it used to be. It was really pretty easy when everything was here and then I backed it up to here. And then I knew where things were.

                                    So, as things go out, they bring a lot of value, but then, there's a ton of loss of visibility, I think, is one of the words kind of used in there. And so, with that loss of visibility becomes a loss of control or feeling, at least because, and what I hear all of you saying is that there's a renewed focus, but especially what AIOps is about is really focusing not on whether something's up or down, because yes, that's important, but it doesn't have anything to do with what that end user experience is.

                                    And the end user experience has so many things that are kind of leaning on it that could have an effect on it. And so AIOps, and so, correct me if I'm wrong, but seeks to say, "What are all the different ways in which you can say, 'What's affecting this end user experience and how do we make sure it's happening in a way that the end user remains happy, productive, making money, whatever it needs to be?'" Is that roughly it? I feel like my words are failing me again. I'm camera-shy.

Tanner Bechtel:            No, I think you're spot-on, I think you're spot-on.

Robb Boyd:                   Well, okay. So you guys had mentioned in terms of kind of a reference architecture that you were building, and I don't know how far we go with this, because I figured something is happening all the time. You guys provide the ability to help customers who are struggling with these issues. And a lot of what you do is not only with the experts like we have here on the call, but also when people can work with your facility, you guys are testing the different vendors' promises, seeing what works well together, really taking all the little silos, as I like to think of it, from the vendors and saying, "How do we make these silos work better together," because that's what the customers are trying to do. So you're doing it on behalf of your customers.

                                    But you'd mentioned, and this was a new one to me in terms of AIOps, but it makes sense. But everything from, of course, a ThousandEyes to Cisco Intersight doing stuff I wasn't aware of, also doing ... What was the other one? AppDynamics.

Tanner Bechtel:            CWOM.

Robb Boyd:                   And CWOM.

Tanner Bechtel:            AppDynamics. Of course.

Robb Boyd:                   Yeah, and CWOM, but let's get back to, because ThousandEyes was one of the ones that, it didn't surprise me because, for what they do, but I like the way this represents something maybe we were blind to it. I didn't realize we'd be blind to on this one.

                                    Neil, I wonder if you could describe for anybody who's not familiar with ThousandEyes who they are, what they do, and why this becomes maybe a really critical ingredient for a successful AIOps.

Neil Anderson:              Yeah. Essentially, if you think about connectivity, it's gotten more complex and often, let's say we have an app slow down. In the past, the first thing is, "Well, it must be the network." So I go and I crawl through logs on my network and try to figure out what's wrong, but often the problem is not on my network anymore. It might be a service provider out there between me and the application workload, or it might be a SaaS provider that is having a slowdown for some reason, but I don't have a lot of visibility into those with the traditional tools.

                                    What ThousandEyes has solved is really putting agents out in all those different locations. So now I can kind of crowdsource, is the way I look at it, global connectivity. I can tell where the problems are. And that helps me with tremendous visibility, because now, if I determine it's not on my network, I can focus on where the problem really is. It's between the user and that workload. So just tremendous amount of insight that's available there that can really help to pinpoint where connectivity problems are experiencing.

                                    And to your point, they're often not, it's not an up and down anymore. It's a slow down. It's it's packet loss or latency or something. And the ThousandEyes tool, we like about it is it gives you just a ton of visibility into that all the way from the user to the app workload.

Robb Boyd:                   Yeah. I always like hearing from ThousandEyes in terms of they always are consistent about putting out information of what they're seeing from all these agents that they have, and kind of reminding us that the internet is literally just kind of a loose collection of destinations all tied together with BGP in a way that mostly works, but is also ripe for everything from routing issues to other things that can have a cascading effect. And we don't know what to blame. In fact, it's less about blame, but it's also about knowing what to fix quickly so that we can resolve things and continue to do what we're there to do.

                                    But there's some other ones mentioned in here, Chris, I wonder if you could reflect on what is the role AppD plays and we'll get into CWOM and Intersight as well.

Chris Weis:                   Yeah. So, I think all three of these things are starting to come together and we'll get into that a little bit, but maybe Tanner will let you cover AppD first and then I can jump into the Intersight.

Robb Boyd:                   Feel free, Tanner. Give us AppD, AppDynamics.

Tanner Bechtel:            Absolutely. Yeah. AppDynamics is the official ... One of the things that we've been able to do, Neil said something there that it's not binary anymore. It's not on or off. It's the slowdown and figuring out where that stuff occurs because we have, everybody says this like hyperbole, like the complexity has exploded, right?

Robb Boyd:                   Right.

Tanner Bechtel:            The Kubernetes apps and containers and serverless models and cloud and IoT and all of these things that innovation leapfrogs. So you get innovation with these environments and what you have to do in responses is come back and figure out how to manage them and effectively deliver them.

                                    And so what AppD is doing is it's focusing on the end user all the way out to you. If you're booking a ticket on an airline website, that experience that you're having. So two things have happened in the tech industry, if I can be top-down for a second,-

Robb Boyd:                   Yes. [crosstalk 00:12:41].

Tanner Bechtel:            ... that have been amazing, that has literally led to our careers, is that we have both depersonalized and personalized almost asynchronously, but in the same space.

                                    So, what we've done is we've taken a simple app model. When I started building software, it lived in a building I could go visit. It was inside of a data center and typically it was inside a one rack. I could open the rack. There's my data server. Here's my web server. Here's my router.

Robb Boyd:                   [crosstalk 00:13:09]. Yeah.

Tanner Bechtel:            Everything's right there. Everything is all inside of it. It has exploded in complexity. And so all that stuff lives in different places. And depending on the time of day, which Chris will describe it like CWOM, it might move around. It might not even exist in the same physical space. So, what we have had to do is we've had to begin to truly understand the user experience.

                                    So, AppD looks at the end user monitoring, like what am I, Tanner? I'm a platinum member on an airline. What am I experiencing in this moment? And I'm translating that end user experience, and I'm using AI or machine learning and I'm actually building models to say, "This group. This is your high-dollar group. Take care of these people. They've experienced a 17% slowdown and here's the workflow and the journey that all of them are taking and right here is the yellow spot."

                                    So, when I double click into that, AppD allows me to say, "Okay, now I can see which part of the application or where in the data or where in the network?" And AppD has a natural limitation for the application.

                                    So, we hit a wall. So, as we talk about these toolsets, ThousandEyes, CWOM, I will eventually all, a myriad of different tools, what they allow me to do is keep that visual, that personalized perspective on the performance of the individual and the application, but continue to dig down.

                                    So AppD right now gives me that application workflow, but what these tools, Neil's and Chris's space, amongst many others inside of this AIOps model, allow us to do is step further and further and further in. And as we do that, we collect analytics, but the big picture, the next step beyond that is if these decisions and these architectures and dependencies of failure continue to be booked and archived and put in the library, we reach a point where we can start to actually apply some of that AI and ML and start to figure some of these solutions out.

                                    So, I go back to the statement I made, isolating high-value decisions. Instead of me being presented 40 alerts from ThousandEyes and AppD and CWOM and Moogsoft and 15 other platforms, I'm given a situation. "Here's what we think happened. Here's what we think it looked like before. We've seen three of these. This has a 92% chance of being the solution. Do you want to execute the fix?" And eventually, we don't have to execute the fix. We focus on the quality of the experience in the app, not the solution.

Robb Boyd:                   Sometimes, and this sounds very similar in my overgeneralization, when you talk about AppD and ThousandEyes is there's a little bit of a pulling back of the curtain where you go, "Oh, that's what's happening." There's a demystification. It feels like magic at first. Then when something goes wrong and you go, "Well, I need to see what's actually happening here." And so you need a way to then go in and have that increased visibility down to whatever minute level is needed to be able to do that. And it feels like it's even more important these days, because as we move to, as we were pushing, everyone's moving into containerization and apps are getting disaggregated or elements or the database is over here backed up with another app over here and these containers are expanding and contracting.

                                    So these are the kinds of tools that we've got to be looking at, regardless. This is not a negative thing. It's just dealing with the reality as it currently exists.

                                    Well, Chris, you punted that back to Tanner on AppD, but so can you tell us more about, first of all, CWOM? Tell us what is that acronym. Just make sure we understand that is an acronym.

Chris Weis:                   Yeah. So, there's a couple of things I'll talk about here. One is CWOM, which is Cisco's Workload Optimization Manager. Another is Cisco Intersight, which is their cloud-based infrastructure management tool. And so, I'll talk a little bit about how those two things are working together today and maybe how they'll work together tomorrow as well.

                                    So, I do find it helpful sometimes to step back and just talk through what's the problem we're trying to solve here and what would be the optimal outcome. And then we could talk about how Cisco is stitching those things together.

                                    So, I think the problem that we see today and Neil referenced it before, but if I'm working on a help desk in any company today, inevitably somebody on the help desk is going to get a phone call that says, "Hey, I'm on the mobile app right now. I'm trying to order a pizza or a car or whatever it might be, and it's slow, or it's not working, or it's getting connectivity errors," or something like that.

                                    To try to debug that issue just sitting at a help desk, wherever you might be, home today, obviously in the times we're in, it's almost impossible because that could be an infrastructure issue, an application issue, a process issue, a network issue. Any of the cloud providers can be having issues right now in any layer of their stacks. It's almost impossible manually to figure that out today. It's too complicated. So, what we'd like to get to is to a point where the help desk user's not involved anymore. It's really all handled by technology and by intelligence in software.

                                    So when something like that, like the application stops or starts responding incorrectly, we want it to be identifiable by software, and we want the software to make an intelligent decision about what should happen to try to resolve that on its own so that, hopefully, these things are identified and fixed before hopefully any users see it, but if they do hopefully just a few number, a low number of users. So, that's like the pie in the sky. That'd be awesome if that's how it all worked.

                                    So, the reality is that it takes some things to get to that point. So, one, we have to be able to see all these different data points. So we have to know what's going on in the cloud and know what's going on in our own data centers and know what's going on in the networking and the storage layers and in the application code. And so this is where things like statistical AppDynamics that will look at the application layer and that will tell us very specifically, what's going on with the code layer.

                                    Something like Cisco Intersight looks at infrastructure, so that will look at things like servers and virtual machines, containers. And it will tell us what's going on at that layer. ThousandEyes will look at networking, and so that will start telling us what's going on at that layer. So, if we can centralize all these things together into one location, which is what Cisco Intersight is starting to do now and we're going to keep moving down that path, then all you need to do is apply something like some data analytics and a decision engine to it, which is what things like Cisco Workload Optimization Manager do. And suddenly, we have all the pieces that we can actually build this tool to start discovering and reacting to issues as they occur.

Robb Boyd:                   Well, you're hitting on something there that I want to make sure ... And so it started to sound like, and I don't think this is what you meant, and I probably just misheard it, but you're not saying Cisco Intersight becomes AIOps by integrating all this information necessarily, because if there's anything I've learned from you guys is that I don't think there's any situation where the one tool answers all problems, but because it's really about how these things are working together now, because there are many customers who have maybe one or two of these things, or maybe they have one and they're looking at one or two others. That may be solving problems that they have.

                                    And I don't know who to direct this to, but I feel like how do you begin bringing the unique information and capabilities that each of these things that we've kind of talked about here represent? How do you begin bringing them together, so that you could have that mythical all-in-one dashboard of omniscient knowledge?

Tanner Bechtel:            That's amazing. Omniscient dashboard of whatever-

Robb Boyd:                   Yeah. You say omniscient again.

Tanner Bechtel:            ... the hell that statement you just made was, that was great. Yeah.

Robb Boyd:                   I don't know what it was, either.

Tanner Bechtel:            I'd say that's what we're seeking. And I can answer this. I think all of us could answer this in our own way to some degree, but I'll take a shot at it first. The main thing that we do and the thing that we have done that has made us unique is that we are not a platform developer. We're not a software developer. We don't create a tool. We maintain a level of, we're based in Missouri, the "Show-Me State". Now, I think we take that pretty seriously. We use the ATC to really prove out the best models.

Robb Boyd:                   It's the Advanced Technology Center there in St. Louis?

Tanner Bechtel:            Yeah, absolutely. Yeah. The Advanced Technology Center, which is essentially our R&D enterprise data center, and we have built models designed around solving problems. So, the main thing that we do differently, and the main thing that any business that looks at this has to think through is that you are solving a business objective. That's what we do when we look at AIOps. Anybody, and AIOps has a term is one of the more confusing things that anybody will ever step off into, because you've got networking teams, you've got software teams, you've got services, groups and management consultants. Everybody will focus on this area. Ultimately, what you have to do is look at the business objective. You have to look at what are you trying to solve, better experience for a customer, decreased cost of operation and start there.

                                    The second part, it diverts at that point. Two primary means of engagement to solve an AIOps problem. The first is product-centric. So, if we go in and we look at tool sets and we say, "AppD exists and ThousandEyes is there, but you don't have CWOM. Let's build interconnectivity and interoperability between AppD, let's build it between ThousandEyes, and let's add CWOM to the mix to provide a richer experience for you to make decisions inside your preexisting enterprise monitoring category."

                                    The second way to do this, which is the way that we are really growing to lean on heavily now, which is, in my opinion, the better way. And that is really to start looking at this from we execute it through an operational readiness assessment. And what that is, it's a series of small workshops that go to the business objective. We talk to the business, we execute small engagements with the right people in the right teams that help us answer what are we trying to solve, what are the areas that we need to fix, what are the preexisting tools that exist, what are their adoption levels, and how can we actually stitch together a journey to get them from today to interoperability and an AI-based model for decision-making?

                                    It's this thing we keep talking about it, we keep kicking it around is very complicated. It has been for us, we've executed it. We've built an MVP. We've actually created not just this thought leadership, but a demo. We can show you. You can log into the ATC and we can walk you through what We Sell Tea. We have this funny little application where you can-

Robb Boyd:                   What? Sell tea? What are you saying?

Tanner Bechtel:            I have to explain this. So, we have built this absolutely beautiful complex model that integrates all of these tool sets. I mean a dozen different platforms for monitoring and network performance and cloud performance and maintenance and management. This complex app that's Kubernetes-based that is sitting on multiple cloud providers and our internal infrastructure. And we're using that giant horsepower monster to run this funny little application that you can go in and you can buy loose leaf tea. It's just an example that we've built for-

Robb Boyd:                   Okay. Oh. I got you.

Tanner Bechtel:            ... showcasing this interconnectivity. But we did that on purpose, and that is because we have gone down this path of complexity to be able to deliver it simply. People start today. People can sit through a webinar and they can sit through a pitch and they can see where data science proves the model out, and that it's a good investment long-term, but the answer to a question that everyone asks, it needs to be where you begin today. And we get that question a lot. How do I start? How do I begin this conversation? Where do I stick the first pin, right?

Robb Boyd:                   Yeah.

Tanner Bechtel:            And so we have sought to do that in an advanced way with an advanced set of technology, but also in a way that we understand that getting started is half the battle, where you began and how do you start proving the ROI? Because it is a long journey. To build an AI-based IT operation structure is not something you do overnight, and it's not something you do with one tool. It's not one platform. It's not one services group. It is a journey you take with a trusted partner. And that's what we have. We've been on this call multiple times in different capacities, but what we have done even internally is to partner up with our own teams to say, "Real-world execution, what's the model? How do we help people step into a zero-entry pool instead of being able to asking them to jump in the deep end without knowing if they can swim or not?"

                                    So, that's a very wordy answer to a short question. I would love Neil and even Chris to add to that too, but that's my take on it.

Robb Boyd:                   Yeah. Neil, I'm curious. One thought that occurred to me also was earlier, not necessarily from anything you guys have said, but just the notion of AI and the fact that AI being a part of the terminology, and I wonder if people get automatically turned off because that's in our buzzword bingo of what vendors going to use AI and ML. ML's probably in there somewhere. In fact, I think I heard Tanner say ML in there somewhere.

Tanner Bechtel:            I did. I did.

Robb Boyd:                   And obviously these are good technologies behind this, but they're used so loosely to mean so many different things. I'm curious. From your perspective, I'm assuming it's not a legitimate excuse to ignore something that's going on here.

Neil Anderson:              Yeah. It's funny. I usually have a joke that I say. If you put the words AI, ML and analytics in a business proposal, you can get VC money just raining it on you.

Robb Boyd:                   That's the formula. Okay.

Neil Anderson:              But because, in reality, there are just a ton of different tools out there, right?

Robb Boyd:                   Yeah.

Neil Anderson:              And everybody is, I think, overusing the terms AI and ML often. They have specific meanings. And I think where I see it in my space is that we're now starting to correlate large data sets of information, where somebody used to go into a log or multiple logs on devices, and they correlated that information manually. We're now seeing automatic correlation of those large data sets and the data sets themselves just expanding rapidly. And you need, it's really beyond the point where a human can correlate a lot of those things is the way that I look at it.

Robb Boyd:                   It turns too fast. Yeah.

Neil Anderson:              And so bringing those islands of analytics together into something meaningful that is describing, to Tanner's point, this situation is happening. It's not about, "Hey, there's a CPU over here on this router that is high. Hey, there's something happening over here."

                                    It's about, there's an app slowdown. Well, now let's figure out what's, it's a sort of that. And I think that there's so much data being created that it's really becoming impossible to correlate it from a human's point of view. And, to me, that's really where the AI applied here and the machine learning takes that to the next step, which is, "Okay. If I figured out that a problem happened this way, and I solved it in this manner before, can I repeat that behavior if I see the same problem again?" And that's where I think the learning comes in.

Robb Boyd:                   The gold standard, repeatable behavior. It's always so hard to find that when you're troubleshooting to make something happen twice, but it also reminds me, and I can't remember if it was in discussion with you guys or not. But this notion that we build networks and architectures to be relatively self-healing or to work their way around problems.

                                    And so, our old mental metrics that a lot of us that have been around a while may go off of is when things are operating, you can't assume if you don't have the visibility that everything is okay, because there could be that next disaster building in the network and parts of your infrastructure could be wanting to give you signals. There are things that need to be addressed by working around, because I feel like it, the goal is to understand how is everything operating so you can optimize because Tanner, you would always drill this into me in previous conversations, it's really about optimization, not about chasing issues, but it's about optimizing, which means we're not waiting for something to fail. You're looking to make everything operate at its level best as the teamwork, but that is so that we can have enough faith in what's happening to then allow some automated things to occur.

                                    So not just, as Neil, as you're saying, in terms of automatically gathering the data, which is important, but also automatically responding to what the data has revealed.

                                    So, if you have good data, you're processing it appropriately, you hopefully will get to a point where some of the decisions that are taking up way too much human capital can be executed upon by the system because it's been tweaked to that level.

                                    And Chris, I'm curious just as we're kind of wrapping up in the end here though, isn't that kind of what you're talking about in terms of some of the stuff Cisco's been doing that you guys are playing with with Intersight and some other things that they've really been building towards this ability to take this data and start making decisions about maybe it's where workloads are running? Maybe it's what's spun up, what needs to be spun down, that type of thing?

Chris Weis:                   Yeah. So you mentioned something before that's kind of near and dear to our heart. We tend to think about ops in general, not just AIOps, but ops in general as responding to some sort of a negative event. And that's true. That's a big part of it, but there's also a flip side of that coin, especially in a multi-cloud world, which is, we want to make sure that we're running the right kind of workloads on the right kinds of platforms to optimize efficiency. And that usually results in cost.

                                    So, whether it's in our own data center or whether it's out in AWS or Azure or Google or something like, that all cost money to run things. Electricity cost money, data, data center space costs money, servers cost money, the billing models and the cloud providers, they all cost money.

Robb Boyd:                   You sound like me yelling at my kids upstairs. "That costs money!" Yeah.

Chris Weis:                   Yeah. Yeah, exactly.

Tanner Bechtel:            Turn the lights off.

Chris Weis:                   So the point is, we want the operational layer to react so that if we need more horsepower, we need more performance, whatever it might be, we could do that very quickly and rapidly and it does great at that. But the folks who care about money, which is everybody who's in this for the business side, that we want to make sure it shrinks back down, too.

                                    So, that's a huge part about what AIOps can do for us. It makes sure that workloads are running in the right place at the right time. And if we don't need all this infrastructure to support them anymore, we can spin that back down or repurpose it for other reasons. And frankly, that's just something that humans don't typically take the time to do today just because they have other jobs to be doing. So, it may cost a little money to get things like AIOps in the door, but the cost savings that come back just out of making your operating model more efficient can easily pay for it.

Robb Boyd:                   Yeah.

Tanner Bechtel:            Yeah. Yeah.

Robb Boyd:                   And it's somewhere we need to go. Sorry, Tanner. I didn't mean to cut you off.

Tanner Bechtel:            Yeah.

Robb Boyd:                   But it's a direction that everything is going that way. The complexity is not going to stop. Things get easier on one side, but yeah, we can't give this. This is not a stick your head in the sand moment. It doesn't feel like, but Tanner, what were you going to say? Sorry.

Tanner Bechtel:            He said something there that I think is really important. It's something that we say a lot, but it takes us a minute to get there because we're talking about the check. AIOps is not a solution that eliminates people. All of our monitoring spaces, everybody thinks that the smarter we get, the more plainer data becomes, that the less people's decision-making is necessary. And it's quite frankly the opposite.

                                    Right now, people are making decisions based on how well they can assimilate the data. What AIOps allows people to do is start making very good decisions based on clear data. It's like being a detective. That's essentially what people do when they are trying to keep these systems up and keep them efficient. They're taking all the data they can collect reasonably, using pattern recognition in their brain, and solving problems as fast as they can get to them.

                                    And so what AIOps allows us to do is to kind of delegate a little bit of that, use pattern recognition at scale, and it's not magic. It's individually designed in applications, event correlation tools and APM tools and NPM tools. It allows them to start making smarter decisions. And I always go back to that. I think I said to you multiple times, "You can't change what you can't measure." You must measure things in order to know if you're doing things better.

                                    That's what we aim to do. The more we can collect and the more we can aggregate and the more we can do that in real time. We did log analysis for years, but we're talking about doing this now in the moment, in real time and using AI as that horsepower to do that in real time. But the answers that come back to smart people. We have tons of smart people in these places that we are enabling to become even more valuable to the institution and quite frankly, making a more sustainable business model.

                                    We can allow businesses to run leaner and smarter and more effectively for the end user. So, we've employed these massive changes to really deliver a more personal experience, which is the first time ever in my technology history I've been able to say something like that.

Robb Boyd:                   Yeah. And I think that's the big thing left unsaid here as we wrap up, but smart people need smart tools and there's no lack of smart people. It's just kind of getting access to the right tools. And to your point you made earlier where people go, "How do I get started?"

                                    I'd like to end with that because you've talked about what you guys have built out. I know you're constantly running the various R&D. There's no shortage of vendors that are always knocking on your doors, wanting to be in your lab. And many are, as you continue to work solutions in and out of what becomes your golden master, to use my own language. I don't know if I'm saying it right, but I wonder if you could tell me what kind of thing could someone be doing to say, "Boy, how do I take the first step in my own operation?"

Tanner Bechtel:            Sure. The one thing we do, and one thing I love about World Wide and I love about working here is that when we learn stuff, we run out in the hallway and we say, "Hey, I figured this out. Everybody, take a look at what we built." And that manifests itself in the platform. If you go to wwt.com, you register, you can get online and you can see everything we've built, thought leadership documents and case studies and labs. You can actually go play with the virtual labs that we use and have used to build out our AIOps model. All of us, all three of us here have different parts of the World Wide business that we manage. All of us have tons of labs.

                                    So, we want customers to go on this journey with us to get started. We're not gatekeepers in this process. We share data, we share knowledge, we share what we've learned and the advancements we've made. Our value to our customers is understanding their business. We understand business objectives, and we maintain a level of agnosticism with tools that allow us to make really smart, strategic decisions informed by incredibly in-depth technology experience. And that's really where we fit.

                                    So step one, go read about our perspective. Go understand who we are on the platform at wwt.com. Go see how we feel about this, how we talk about it in each of our respective areas.

                                    Step two, get in touch with us. Every one of those labs has a way that you can actually communicate with people who either created the lab or lead the division or lead the technology space. In our case, we will walk you through our entire AIOps infrastructure, real time, real demos, not PowerPoint slides, led by real solutions architects.

                                    And the third part of that is to engage with us at a consultative level. We can actually start to help you down that path whether you're beginning in networking, you're beginning in multi-cloud, or you're beginning at the app level, or where have you, wherever you live in the enterprise. We have teams of people that have designed programs to help people move through this journey. It's not a tool. It's not a product. It's not a single solution. It's a journey that you take with us. We share everything we can in the process to get you there because we love mutually intelligent clients. That's really where it lands.

Robb Boyd:                   Yeah. It does make things easier when your clients are smart, too. And I love the fact that before we were all corona-ed, you guys were already making the ATC available online well before we all got stuck at home. And, so that obviously continues today as you continue to invest in it.

                                    And so we're not saying anybody needs to travel to St. Louis or anything like that. These are all things that are available remotely. And, the ones that I've sat in on and such, I do like the sharing and the fact that you guys are going through a lot of the same things that you put your customers through, of course. And so, what you do is you share kind of the learning behind that, and the mistakes that we may not talk about on a platform like this per se too broadly, but one-on-one with a customer who's experiencing the same things.

                                    There's an empathy there that translates into an intelligent understanding of how to make this go faster so that you can avoid repeating the same mistakes over and over again. And one of my favorite leverage, some of the customers need to be looking at you guys, I think, to leverage stuff that they've already invested in. It's not always about buying something new. It's really about getting a lot of value out of what you already have. And I think most people like me, just personally, I own way more things than I know how to actually operate, to a degree of ... Because the difference like Tanner, you've got guitars behind you that I know you know how to play. I've got guitars behind me I don't know how to play, just to show you how shallow I am, but no, you guys are rock stars and I like what you're doing and the methodologies behind it that you can access.

                                    So, thank you, Chris, Neil, Tanner as always, of course, and World Wide Technology, wwt.com. You guys need to join that platform actually, because it makes it easier to subscribe to the stuff that is published and, of course, engage in the labs and all the materials and resources that they continue to make available because it's more of a community than a broadcasting platform by all means. So please reach out and do that.

                                    Thank you so much for watching. This has been the TEC37 podcast. My name is Robb Boyd. We'll see you on the next episode.

 

Comments