 Live from San Diego, California, it's theCUBE. Covering Cisco Live US 2019. Brought to you by Cisco and its ecosystem partners. Welcome back to San Diego, everybody. You're watching theCUBE, the leader in live tech coverage. My name is Dave Vellante, and I'm with my co-host Stu Miniman. This is day two for Cisco Live 2019. We're in the DevNet zone. Stu, I was walking around earlier in the last interview, and I think I saw Ron Burgundy out there, but so. Stay classy, San Diego. He said, all right. Ling Ping Gao is here. He's the founder and CEO of NetBrain Technologies just outside of Boston. Ling Ping, thanks very much for coming on theCUBE. Thank you, Dave. So, you're very welcome. So, I want to ask you, I always ask founders, your passion for starting companies, why did you start NetBrain? Well, maybe tired of doing things manual. Well, that's one side. The other side of it is I used to took an exam called CCIE, a lot of folks doing here. I failed my first try. It was a big blow to my ego, so I decided to say, we're going to create a software, help them to pass. This is actually the genesis of NetBrain, help people to be better on doing their network management work. That's a great story. So, tell us more about NetBrain. What do you guys know all about? Sure, we are the industry's first just-in-time network automation software. Our mission is to democratize network automation. Every engineer, every task, they should have started with automation before human being touched the task. Yeah, you know, if we go back, let's say 10 years ago, people were afraid of automation. You know, they thought it was going to take away their job. They still are. And they still are, we'll talk about that. And I want to ask you about the blockers. And they were fearful, they wanted to touch things, but the reality is, people talk about digital transformation. And it's really all about how you use data, how you leverage data, and you can't be spending your time doing all this stuff that doesn't add value to your business. You have to automate that and move up to more valuable tasks. But so, people are still afraid of automation. Why? What's the blocker there? They have the right reason to be afraid, because so many automation was created once, used it exactly once, right? And then you have the cost of traditional automation. You have the complexity to create a network automation. You guys realize that network automation, you cannot have network automation only work on a portion of your network. You have to work on majority, if not all of your network, right? So that's become very complex, just like you want a self-driving car. You can go buy a Tesla, a new car. You can drive on its own. But if you want your 10-year-old Toyota driving on its own, retrofitted, that's very complex. Well, that's today, network automation had to deal with it. You had to deal with multi-vendor technology, multi-years of technology. So people spend a lot of money. The return are very small. So they have a right to be afraid of them. But the challenge there is, what's the alternative? Well, wait, before you go there. So there, if I understand it, just playing back. They're solving a very narrow problem. They do it once, maybe twice. Maybe a rudimentary example would be a script. Yeah. Right, and then it breaks, or it doesn't, or something else in the network changes. And it really doesn't affect that, right? Yeah, I mean, I think back too many network engineers is like, well, I'm sitting there. I've got all my geek knobs and I get everything done. And they say, no, don't breathe on it because it's just the way I want it. That's right. That can't be, that doesn't scale. It doesn't respond to the business. I need to be able to respond fast to what is needed and things are changing in an ever environment. So it's something that I couldn't as a person or a team keep up with myself. And therefore I need to have more standardized components and I need to have intelligence that can help me to manage my, that's it. So we've laid out the problem. What's the better approach? Well, if you're looking at the challenge today is you have to have DevOps, which a lot of here, DevNet engineer, know how to script. And the NetOps, the engineer who know how Neto operates, work together. So there's a data part of it. There's a knowledge part of it. This too has to meet to create a network automation and that network automation has to be a scale. So the challenge of traditional network automation, why is for sure lie on, if you're going down technical level, is one is the data, too many data, unstructured. And the other one is on knowledge. The knowledge cannot be codified. So you have the knowledge sitting people's head, right. A programmer have to work on with the network engineer together to start. You make the cost higher, you make it very unscalable. So those are the challenge. So how fast forward will we have to do? So Neto brand for the last 15 years to decide to look at differently. So we created something called operating system of total network. And actually use this to manage over a thousand vendor models technology. And through platform, you can continually adding new things into this platform. So the benefit of it is the network engineer, anybody can create automation. They don't have to know how to writing a code, right. And DevOps who know the code can also use this platform. All the people who are familiar with technology like Ansible, they can integrate that to Neto brand. Okay, so you have all this data, which I can say is unstructured, so it doesn't have any meaning. Data is plentiful, insights aren't, right. And then you have this what I call tribal knowledge. Joe knows how to do it, but nobody else knows how to do it. So you're marrying those two. How are you doing that? Are you using machine intelligence and then iterating, building models? Can you add some more color as to how you go about that? What's the secret sauce? We took a hybrid approach. First of all, you have to model the entire network. With this is called operating system of total network. We have about 12,000 valuables modeling a device. And that 12,000 valuable as to go across your, let's say a thousand node network, there will be 12 million valuables describing your network. That's first. Then on top of it, the 12 million valuables will be continually monitored through AI and machine learning, give something called a baseline data. But on top of it, the user, the human being will have the knowledge on what is considered normal, what is considered abnormal. They can add their intelligence through something called a executable run book on top of this system. And that system now can be run at any time, which talking about when somebody attacking you, when network is on a fault, or you threw a human being on a task, now the automation can be run just in time. So the expert, the subject matter expert, the domain expert, the person with the knowledge, he or she can inject that knowledge into your system and then it iterates and improves over time. That's right. And it iterates and improves it. And other people can open the hood, can continually improve the intelligence. So the whole automation in the past why is the writer once only used once because it's a closed loop. It's a script. The UI, input and output are just text. So it wasn't designed with a comprehensive model behind it. So you do it, you're building your model, you're writing your logic within a same period of code. We decide, we think that's, you cannot scale that way. Okay, so obviously you can stop Dave from inputting his lack of knowledge into the system with security control and access control. But there must be a bell curve in terms of the quality of the knowledge that goes into the system. You know, Joe might be a superstar and Stu maybe doesn't know as much about it, no offense Stu. Stu knows good. So how do you sort of balance that out? Do you try to reach an equilibrium or can you weight Joe's knowledge more than Stu's knowledge? How does that work? So the idea that this automation platform has something called executable Rumble. That executable Rumble can be shared and collectively improved by three sources. One is engineer themselves, right? The other one is by underlying engine. So we talk about AI and machine learning, we have that, we also have a low engine. We basically are adjusting that ourselves. Certainly it's through a collaborative partner, for example Cisco, who run many years of hardware. They have a lot of know-hows, they attack. That knowledge can be pushed to the user. We actually have a, in our system a partnership with Cisco Tech. Southern though script can be wrong, slow native brand without a user involvement and getting the benefit of, without talking with tech, getting the answer. Yeah, so I think you actually partially answer the question I have is how do you make sure we don't automate a bad process? Yeah. So and maybe talk a little bit about kind of the training process too. Your original why of the company is to make things easier. You know, what's the ramp up period for someone that gets in, give it a little bit as to how many engineers you guys have worked with? Well the ultimate goal, I mentioned our mission statement of native brand is to democratize network automation. You know, used to be network automation, only the gurus-gurus do it, right? DevOps and et cetera. And young generation, my generation who use command line, this is not us, right? This is the same, you know. But we believe nowadays with the complexity of NATO, with the cloud computing, with the cyber security demand, the alternative to network automation is just no longer viable. So we really put a lot of thought into it. It's like how we can put network automation into everyone's hand. So the things we did as three angle of it, one is automation can be created by anyone. The second meaning developer of NATO, of anyone who have knowledge of NATO can create automation. Second piece of it, automation can launch at any time. Somebody attacking you in the middle of the night, they don't tell you, automation can launch to protect you. A network is out, you don't have people are dying on the job. Automation can launch the diagnosis. So it's called a launch at any time. Third one is can adapt to any workflow. You have troubleshooting, you have network changes, you have compliance, right? You have documentation workflow. The automation should be able to adapt to any of this workflow. The top integration, for example, we have with service now. So there's a ticket. Human being shouldn't touch the ticket. Before automation has done its share, right? Then human should come in and then continually use automation. So you talk about democratizing automation, network automation, so it's, so anybody who sees a manual process that's wasting time can sort of solve that problem. Is essentially what you're doing. That's what we're doing. So is there a pattern emerging in terms of best practice, in terms of how customers are adopting your technology? Yes, now we see more and more customer creating these things almost like a club. The power user, and we often call it normal user, they have the knowledge in their head. The pattern emerging we saw is they now work proactively say, how can I put that knowledge into a set of excusable format so that I don't get escalated all the time, right? So that I can do the things more meaningful to me, that I've been repeating the same thing 10 times a month, right? And I should one of my, we call it a shift to the left. Level one doing level two, the machine doing the level one's task, level two, level three are doing more meaningful things. How different is what you're doing at NetBrain from what others are doing in the marketplace? What's the differentiation? How do you compete? Yeah, little going to mention so far has been a piecemeal. I think a fragment, it's things that has done, typically in a three big category. One is wholesale on the hardware approach. You replace the hardware with SDN, SDY, SDLAT. There's automation capability building to it. I call it a Tesla approach. Buy a Tesla and you can drive and self-driving. The second approach is software approach. This is where we are leading. Build a model of your entire network, apply machine learning and statistical analysis behind, but also more importantly, open architecture. Allow a human being to put their intelligence into this. That's the second approach. And the third approach is actually service. Let an outsourcer help you or moving your workload into cloud because there's a better automation there. So we are focusing on the middle portion of it and the landscape is really well. We have over 2,000 enterprise customers and they are automating, this is not just one or twice a week, but 1,000 times a day. We're really excited that the automation in that scale is transforming how little is being managed and enable things like collaboration. There used to be people from here, people from offshore couldn't work together because data and knowledge is hard to communicate. With automation, we see collaborations happening, more collaborations happening. We've been talking about automation in the network for my entire career. It feels like the promise has been there for decades. That's right. It feels like over the last couple of years, we've really seen automation, not just in networking, but we've been covering a lot like the robotic process automation, all the different pieces of IT are seeing automation bring in. Gives a little bit look forward. What do you predict is going to happen with automation and IT over the next couple of years? In the future, that's great. We have cloud computing, we have cyber security, we have the sheer scale of network, driving the network automation to the front and center of the solution. And my prediction in the next five years, probably surrounding one, I think automation going to be ubiquitous, going to be everywhere. No human being should touch a ticket without automation to the first, to the test first. Second, we believe things called a collaborative network automation will be happening. In other words, network automation is following the packet from one network entity to the other entity. The example would have been your managed service provider and enterprise. They are collaborative manager network or common network. But when there's something wrong, we don't know which part have issues. So automation defined by one entity could be run across multiple. Service provider like cloud provider also have automation can be initiated by the enterprise client. We also see the hardware vendor like Cisco and their customer has collaborative automation happening. So the next five years will be very interesting. The manual way to manage and operate network will be finally go away. Like Rick, last question. Give us the business update. You mentioned 2,000 customers, you're hundreds of employees. Any other business metrics you can share with us? Where do you want to take this company? We really want to behind every enterprise. Well, our mission is a democratized network automation. We looking at it in the next five years, our business should have grown 10 times. Well, good luck. Thank you. Thanks very much for coming on theCUBE. Great story. Thank you. Thank you for the congratulations for all your success. Thank you. Keep it right there, everybody. Stu and I will be back. Lisa Martin as well is here with our next guest. Live from Cisco Live 2019 in San Diego, you're watching theCUBE. We'll be right back.