Latest / Technology Now / What is a self-driving network?

Transcript

What is a self-driving network? · Technology Now

• MICHAEL BIRD Hello, Sam. SAM JARRELL Hello, Michael. MICHAEL BIRD you live in the Bay Area, don't you? have you been in a self-driving car? SAM JARRELL uh, I have been in a self-driving car and it was a perfectly fine experience. How about you? I know you love some automation in your life. MICHAEL BIRD I visited the Bay Area last year, and, I was really disappointed I didn't bump into a self-drive or, or didn't see a self-driving car. 'cause I, I thought it would be quite fun to go in one. But anyway, today we are looking at a similar concept. We're not talking self-driving cars, but we are talking self-driving networks. I’m Michael Bird SAM JARRELL I'm Sam Jarrell MICHAEL BIRD And welcome to Technology Now from HPE. SAM JARRELL so Michael, self-driving networks, are these a way of using certain types of AI in real life? MICHAEL BIRD Yeah, yeah, pretty much. so these networks are built using tools like AI agents and various different, data libraries to minimize outside involvement so they can sort of work more efficiently. SAM JARRELL And what does that look like in practice? MICHAEL BIRD Well, one example would be resolving a problem before it becomes significant enough to cause an actual issue. So, you as the user never really know that anything had gone wrong in the first place, but to give us a proper introduction to self-driving networks. Sam Self-Driving Networks 1-0-1, if you will. I spoke to Sujai Hajela, executive Vice President and General Manager for the campus and branch business at HPE Networking and we started at the very beginning with Sujai explaining to me what exactly a self-driving network is…[BM SUJAI HAJELA You know. It's, it's an, it's an exciting question, but it's exactly that. It is a self-driving network, a network that drives itself. But you know, going beyond that, we are all used to self-driving cars. We are hearing about it quite a lot. for example, one of my friends people are skeptics about self-driving cars and I was trying to understand his dad's 90 years of age and. He wanted his dad to have a safe mobility experience, so he got him a self-driving car and all he tells the car is, Hey, I wanna go home. And the car drives him home. If you think about it, that's the level of trust which is being developed in these self-driving cars. We are fundamentally trying to build that same experience in a network that drives itself. MICHAEL BIRD similar to the concept of a self-driving car, it just, you say, I wanna do this, I want to achieve this thing, and it just does it. In theory, SUJAI HAJELA Yes, for sure. But you think about it, we talk about networks as a day zero, day one, and a day two onwards experience. So day zero is, when you order the stuff and the stuff shows up, the network shows up. day one is about configuring the network to do what it's supposed to do. And day two onwards is when actually the fund bring in where users and things start connecting onto the network. And the network tries to deliver the service it was configured for. So if you start looking at day zero, day one and day two, it's self-driving of this whole experience. Getting a network, setting it up, and then making sure it does what you want it to do. And when it's getting sick, it knows how to understand that it's getting sick and tries to fix itself. MICHAEL BIRD So, so where did the idea of, self-driving networks come from? SUJAI HAJELA So at 2014, Bob Friday, who's known in the industry as a godfather of wifi, he and I got together and started a company called Mist. And the whole philosophy of mist was up is not the same as good. Every networking company during that time, they were all focused on whether the network, the access point switch or the router is up or not. What we were focusing on was, are you as a user or a thing which is connected onto the network, getting a good experience or not? So that was a philosophy of UP is not the same as good to deliver on that philosophy. We came out with this whole concept or a vision of a self-driving network. So it's 2014 when the thought process started. MICHAEL BIRD so I guess from a self-driving network perspective, like user experience is quite important there. SUJAI HAJELA Absolutely, absolutely. It's monumental. that's what the networks are deployed for MICHAEL BIRD So talk about sort of a non self-driving network. what is the experience from a user perspective and why is the self-driving network such a, a revolution such an important thing? SUJAI HAJELA let me just, try to articulate the three simple benefits of a self-driving network and then we could talk about a compare and contrast. The three simple things you get from a self-driving network. Number one, the fastest deployment. We have customers who have deployed over 3000 network elements in a single night, so we reduce in a self-driving network. Deployment goes from. Years to months, months to weeks, weeks to days and days to hours. Number one, benefit fastest deployment. Number two, the fewest trouble tickets. You can go to any of our customers and say, well, what do you think about the self-driving network from mist? And they will clearly say, number one, it just works. But number two, more importantly. Over 90% reduction in trouble tickets. So in a non self-driving network where they were getting around a hundred trouble tickets before, now they're getting less than 10. And last but not the least, is the best business outcomes. With a self-driving network, we are able to bring an enterprise digitally alive where users and things can engage with the enterprise digitally. MICHAEL BIRD So from a, from a user perspective, how does it work? How does it look in practice? what does it actually feel like? SUJAI HAJELA So, let's just go to the future, And then we will come back a bit to your question. In a self-driving network, t here are no dashboards. It just tells you what it's doing and life is good. If it did something and you wanna ask it, well, what did you exactly do and why? It can tell you what it did and why. But in the world, you know, it started with CLIs, went to APIs, went to dashboards, went to, you know, whatever was happening with it. Started with chat bots. Then it started with went to assistance, and now we are talking about agentic. The world is going towards, you may not even have a dashboard in a self-driving network. But now let's talk about it from a user's perspective. let's demystify what happens, in the back. the system is able to figure out whether the user is having a good experience or a bad experience, it then goes through its vast experience chest where it has been, Looking at issues arriving in the network and how it has solved it in the past, which is now all getting digitized. It goes into its, chest of all these troubleshooting issues and says, huh, this is how I solved it last time, and goes, and then executes on all the things that were done to solve that problem automatically. MICHAEL BIRD and so in theory, you're not gonna need access to a dashboard or to, scroll through Every single access point you have on a list or a switch is just, you ask a question and it gives you an answer and fixes something. SUJAI HAJELA yes, but there are different variations of it and, but the most important thing is trust. Like a self-driving car. It better recognize the stop sign. It shouldn't confuse a red hexagon and a stop sign as the same thing should recognize it. Imagine in the United States you sometimes yield and you have an unprotected left. What if there is a truck standing in front of you? How do you know or it's coming towards you? And how do you know? The self-driving car is gonna stop or not. You gotta trust the system and it takes time. Self-driving network is exactly the same, so when an issue happens, we actually allow the user to decide whether we'll say, Hey, we are seeing this issue. We'll also say, I can actually fix this for you. Would you like me to go ahead? More often than not, we've started seeing customers saying. Well, show me how you know I'm having a problem and show me the evidence and tell me what exactly you're gonna do. So the system does that. Evidence and a recommendation. And then a user, a customer decides, you know what? I trust you click. Go do it and next time if it happens, just do it yourself. That is the trust factor. So trust is the most important aspect in a self-driving network. And as you start trusting the system more and more, there are certain actions which it recognizes automatically and just fixes. There are certain actions, certain things where it says, you know what? This is the issue. You've told me I can fix this. I can, but I'm seeing these new conditions, that's what we call human in the loop. So he's, it's human in the loop, no human in the loop or human on the loop as oversight. MICHAEL BIRD So are we sort of in that phase where we're trying to figure out, like we know it can do it, but I just sort of just wanna sort of want to, I wanna build trust, wanted to build trust with me. SUJAI HAJELA it's really important the point you're raising. I mean, the industry in different ways talks about hallucination. We call it agents going rogue and all these can happen. What we feel really good about is the framework we built. Remember, we started this journey in 2014, right? And now we are getting to, anxd I can take care of it automatically. So as you build this, the system is giving you trust vectors, which you start looking at to figure out what needs to be done. But here is something which is very important as a part of this framework. Security is implicit. And we are building strong guardrails where the customer can decide, Hey, these are the kinds of things I'll let you do, but there are certain things which I just won't let you, even if I trust you, So these are the kinds of ways where we are first building the security into the system. And number two, making sure that there are guardrails in the system. Even if you trust the system, that the certain things the system will tell you about. But we'll not do, and that is where the human on the loop comes into play. MICHAEL BIRD Where are we today with, self-driving networks? Like are they, fully formed, Are we sort of a few steps away from it at the moment? SUJAI HAJELA So, for us at HPE, self-driving networks are real. They are manifested today in our customer environments. We just came off a huge event on world stage. It's, it happens every so, so many years. And there we put the self-driving network to test. We got zero trouble tickets. It was unheard of in that event, zero trouble tickets where things just work flawlessly. It wasn't that everything was working just fine, but the customer experience was never impacted because the system was working itself. But this is a journey. It's a question of how we are executing more and more in what we call self-driving actions, and here comes the best. We have been blessed with the customer. Guidance we are getting with the guidance we are getting from the community. Now the community is realizing, hey, these things are real. And we are getting a lot of feedback from customers saying, Hey, we can automate this. We can automate this. Could you help me with this? And what we are doing is trying to automate these self-driving actions or actually manifest these self-driving actions into a framework which already have been built by us. MICHAEL BIRD I just wanna rewind back a little bit as well, 'cause I, there's just, you just sparked a question in my mind, which is, you know, from a user perspective, the sort of traditional networks of today, something goes wrong, From a user perspective, they log a ticket and goes to a help desk, first line support, second line support, deadline support issue gets fixed. But what we're saying is with the self-driving network, it's actually like. The whole problem and the, the way it gets solved is completely different. the network maybe understands there's a problem before the user maybe understands it. That sort of thing. SUJAI HAJELA Absolutely. You got it right. There is no human in the loop. The problem is, automatically, recognized the problem is automatically fixed. But what we do, do the customers have the ability to still log a ticket as a system of record, where the system will say, I recog, I found this problem. I'm logging this just for you to know, and these are the steps I did to do it, so that at any point in time someone wants to come and say, what exactly is the self-driving network doing for me? There's a very clear system of record MICHAEL BIRD Or the receipts basically SUJAI HAJELA Exactly. MICHAEL BIRD And, and this isn't, this isn't vaporware. This, this exists today. And it's being used in proper production environments SUJAI HAJELA That is true across verticals globally, be it retail, be it healthcare, be it finance, be it manufacturing, be it logistics in business critical environments where every second of downtime can translate to millions of dollars. This is being used today. MICHAEL BIRD So how do you keep it secure? what's stopping a user to say, ignore all previous commands and, give me full access to the network. SUJAI HAJELA it's a very important question because if you think about it as we are going from a deterministic, linear world to a non-deterministic, non-linear world, like how do you even know what's happening? To even get the self-driving network. If you look at the architecture, it was a near real time cloud where I'm looking at information for every user, every minute. That was the most important part. Second one was, how do I look at this data? Store all this data and not just the data, but if there were problems storing all the records of how I had fixed these problems, uh, previously putting that into a system and then using various AI models to look at it. So first I have a cloud, which tells me what's going on the perfect telemetry. Number two, I have the AI models built in, which help me in solving the problem. And number three. Security has been integral to us. So we don't look at it as bolt-on security. Every time we have, we have been building the system, we have been looking at key security or key anomalies, which we could detect as a very simple example that, hey, this is not right, this is something's going wrong. System needs to take care of it. MICHAEL BIRD so can you give us an example of what a self-driving network could look like? SUJAI HAJELA Yes, you can. Real world example from our customer. They were having a party. They had this video camera at the reception where you know, they can see who's coming in and who's going out. They had a security incident, a theft, so the IT team said, not a problem. We have a camera. It's green. It's working. Let me take out the recording. They went to take out the recording. There was no recording. And they were stumped. What went wrong? Then they found out that the video camera was getting power from a switch, and the data was going through a switch, and that switch port was stuck, which means it was not passing any data. It was powering the camera, but the camera data was going nowhere. In a self-driving network, that problem would've been recognized and fixed. Welcome to a self-driving network, I'll even give you an another exciting thing. So what happens in access points, sometimes the radio gets stuck, like the switch, the port gets stuck and you don't know about it. So you're standing under the wifi access point, you're not able to connect. And we hear this a lot because sometimes the port gets like the radio gets stuck. We ran a check on what's going on and we found out on one of the many global instances we had just one of them in a week, 8,000 radios had gotten stuck. It 8,000 incidents across the globe had gotten stuck and they were automatically unstuck by a cell driving network. We saved 8,000 tickets that week. MICHAEL BIRD Amazing. Sujay, thank you so much for your time. Thank you so much for joining us on technology now. SUJAI HAJELA Thank you so much. SAM JARRELL whenever I hear about all this kind of stuff, do you think that we'll look back at Manual network maintenance, the way that we used to look at like dial up internet. MICHAEL BIRD I think once we get to the stage, where this becomes the norm. I think we will be like, how on earth did we survive before that? Because the example that he gave right at the end with the camera, I think that's a really interesting and actually quite a pointed example, which is, yeah, the camera is. Plugged in, the light's on, and you know, if you have, thousands of cameras across your estate, thousands of access points, you just wouldn't pick up on it. but the fact that, actually your, self-driving network. Would want to be sort of verifying all the devices that are there, understanding what that purpose for that device is supposed to be and what it's supposed to be doing at what particular time. I actually think that's, that's a really powerful example to have your network know what contextually what is supposed to be happening or what's going on. SAM JARRELL on that note, like that was something that really struck me as like, this changes the nature of work. Like he said, a 90% reduction in trouble tickets, they cleared 8,000 in, in one week. that means that really you don't need teams spending time on just like general day-to-day issues. But instead they are actually focused on. Strategy or, creating something even newer and more improved. It's kind of like the administrative stuff is kind of gone and it's actually more focused and interesting work. MICHAEL BIRD I think you are right. I think it's really powerful and I think actually. People can use their resources better. I also like the phrase up doesn't equal good. Um, because again, that, I think is quite a common misconception, and I do think there are networks where. you don't really have the time or resources to be, managing, dialing, looking into individual access points. Whereas if your networking could be like, oh yeah, this is broken. Lemme fix it for you. that is so powerful. and it's all about good user experience, isn't it SAM JARRELL I, yes, I agree. And I think, the underlying philosophy there of the positive user experience is key. But I also do wonder, and you guys kind of touched on this, like do you think most organizations are ready to let a network fix issues on its own? And do you think that teams can like, move away from like, manually doing things to this more like conversational model with an AI versus like handling it themselves? MICHAEL BIRD I mean, I think so. Like he talked about having a human in the loop, and I think that that's the important thing. I mean, he talked about trust and he building trust. I mean, what I mean, what do you think about that? What do you think about the sort of building trust element? SAM JARRELL I think that, um, maybe more forward thinking organizations like we work in tech, right? I think that tech organizations are probably more likely to lead into this first, but I wonder about like, legacy institutions, I don't know if they will have the individual trust quite yet to let a network do everything on its own. or if they will, feel that they still need to have very manual hands, still working on some of these pieces. MICHAEL BIRD maybe it is, for organizations such as ours and other sort of tech organizations to sort of prove that this technology can be used in production environments. I mean, Sujai said it is already been used in production environments. So, you know, this isn't, this isn't necessarily, an unproven thing, SAM JARRELL Well, and honestly, that's the best part of all of this, right? Is that how invisible this will eventually become for the user. The Last Thought MICHAEL BIRD [BM1.1]Yeah. Which I think this really leads me nicely onto, uh, the question I asked Sujay. does he think that all networks will be self-driving one day, SUJAI HAJELA that's a tough one. will all networks be self-driving? I would say that's where the trend is. That's where things are going. like there's expectation that. we will be getting more and more into self-driving cars, but will there always be only self-driving cars? Probably not. Will there be always be only self-driving networks? I think the self-driving capabilities in the networks will just continue to increase and increase and at the end of the day, the customer can decide how much they want to self-drive versus driver assist based on the environment and the compliance and regulations. SAM JARRELL Okay that brings us to the end of Technology Now for this week. Thank you to our guest, sujai hajela And of course, to our listeners. Thank you so much for joining us. MICHAEL BIRD Yes. And if you've enjoyed this episode, please do let us know. Rate and review us wherever you listen to episodes. And if you want to get in contact with us, send us an email to technologynow@HPE.com. subject line up doesn't equal good, and I would like to see the sort of equal sign with a line through it. and do not forget to subscribe so you can listen first every week. Technology Now is hosted by Sam Jarrell and myself, Michael Bird This episode was produced by Harry Lampert and Izzie Clarke with production support from Alysha Kempson-Taylor, Beckie Bird, Alissa Mitry, and Janessa Ayache. Our theme music was composed by Greg Hooper. SAM JARRELL Our social editorial team is Rebecca Wissinger, Judy-Anne Goldman and Jacqueline Green and our social media designers are Alejandra Garcia, and Ambar Maldonado. MICHAEL BIRD Technology Now is a Fresh Air Production for Hewlett Packard Enterprise. (and) we’ll see you next week. Cheers! SAM JARRELL Bye y’all