What Does the Future of NetOps Look Like? John Marchese on ONUG’s Built for Trust

Hi everyone, I'm your host Nick Lippis and welcome to the Built for Trust podcast where you'll get to hear from all the folks who are building and shaping AI enterprise infrastructure. and let's get right into it with our guest. John Marchese. Welcome to the Built for Trust podcast, John. Welcome.

Thank you. Happy to be here, Nick. Really appreciate you having me.

I'm like, I'm sure I've been looking forward to this, this discussion. for two reasons. One is that my wife's grandmother's maiden name is Marchese. So, and I have not ran into a Marchese, another Marchese until you, and she was a wonderful, sweet, adorable woman. So you have, you know, I have high hopes for you. Well, yeah, adorable. Maybe that's the bit. Well, you know, adorable could be placed in lots of different, you know, kind of context, you know, it doesn't have to be looks, you know, so yeah.

Well, what's interesting, like the first test was you knew how to pronounce the last name. Right? So, yeah, oftentimes it's someone who has a relative with a similar name. in order for them to pronounce it correctly.

Yeah, yeah, well, it's a name part of our family. Anyway, so I've been really looking forward to you for that, but also because of what you've been able to create with Network to Code. So let's kind of start there. But I don't want to jump into the company. I want to give you kind of an opportunity. to talk about how and why you started Network to Code, what was the motivation. And also just as a preamble for everyone, like Network to Code is like really kind of a beloved company within the community. And now it's like, you know, these guys do. their next generation network operations and the tools to enable that. And network operations is going through a fundamental transition. So we're going to talk about that during this. So John, really it's to you. how and why, you know, to start it. Okay. –

Great. Great. I'll certainly give you a background on myself, how I got to Network to Code. You know, I started my career in networking back in the late 1990s. I was actually still in college when I was interning for, at the time, Port Authority of New York, New Jersey. And we were… based out of One World Trade Center. And that's when I was first introduced to networking. Excuse me. and It was interesting because I was… Working. on a traffic engineering team within the Port Authority of New York, New Jersey. But there was kind of like a shared responsibility in maintaining the local area networks. And one of the engineers in our group was in charge of the LAN for that department. So I was there just assisting and learning. Then the following summer I worked at Lehman Brothers, which Lehman Brothers had a great networking team, a lot of folks. in the industry came from Lehman Brothers with strong networking experience. And then right out of college, once I graduated, I moved out to San Jose where I worked with Cisco Systems, I worked for Cisco Systems and they had sent out a bunch of recent college graduates to go through a six month networking, training. And that was just an amazing experience. And they were training us on not only Cisco technologies, but the TCP/IP stack and, you know, getting our basic certifications and really… enabling us for, you know, the real world and being systems engineers at Cisco. So when I came back to New York, New Jersey, that area in 2000 is really when I first dove into computer networking.

So what year was that?

Yeah, that was 2000. That was the year 2000.

It's interesting. I know this is all about you, but I do want to just inject a little bit.

Yeah, absolutely. Feel free.

I started working at Digital Equipment Corporation, so I was kind of a decade ahead of you. and at DEC I was in an internship program and it was a two-year internship program. and for literally about, probably pretty similar, about six months of those two years was deep diving into networking. So that was really the kind of like so. 'Cause we were just starting to get into, we were starting to build local area networks and routers were back switches, you know, at the time, probably remember that from Lehman Brothers, you know, we started having the first bridges, you know, coming to the marketplace. And so like, you know, there was all this new fundamental kind of computer networking technology that universities weren't exposed to. And so you needed to kind of train everyone. So I can totally relate, you know, and those. That was really kind of at the forefront of kind of building the economy that we have now with kind of the technical underpinnings of the economy that we're all living in. That's right. I digress. It was a great time.

Yeah, it was certainly a great time and certainly a lot of innovation at the time. I know my early years at Cisco was right when– IP telephony was coming out. There was a lot of movement in the firewall space and Wi-Fi. So it was a great time. So I was a systems engineer for about five years. I went out and– I passed my CCIE exam and lab and was certified in 2004. But I reached a point where I felt it– The SE role was a lot about speeds and feeds. It was, you know, the conversations were, you know, what's the backplane of this supervisor or, you know, throughput and oversubscription. And interestingly enough, I know some of the AI conversations is really bringing back hardware, you know, elements of it and some of those speeds and feeds. But You know, I really, really enjoyed it. I had the opportunity to learn a lot, then moved into a management role. But in 2006, I was tapped on the shoulder by the former vice president of sales at Cisco. And he had just recently left to join a startup organization. It was a VAR called Blue Water Communications Group. And they were looking for someone to build out the engineering organization there. So I made the move to Blue Water and it was just an absolutely amazing experience working with a lot of veterans in the industry and experienced leaders there. And we grew that company from zero to 200 plus million dollars in sales in just a matter of a few short years. And I grew the engineering organization and the company ended up being acquired by Presidio, a national solution provider. And I took on a team in the Northeast area and we ran pre-sales, architecture. service delivery, some managed services components there. We had a staff augmentation business and it was just a great experience to grow from this company that was zero dollars in revenue. Now we're part of a $2 billion plus organization. And it was just a tremendous experience for me. After 10 years in total between those two organizations, it was time for me to take a break. My wife was pregnant with our third little one. And I said, let me get my break in now. And we were able to travel a bit and spend time with the little ones. And since then, we had a fourth. So we have my hands full, I think, and just look. this chaos that surrounds me. But yeah, so we had a great time. Yeah, we had a great time traveling with a family and it allowed me to think about what it was that I wanted to do next. and You know, I realized that looking back at my career, I didn't really, really enjoy working for a startup company, right? Being able to, solve problems the way you see fit, right? Just wearing multiple hats and not just being handed the data sheets that tell you what you need to say or regurgitate or you had to create them, right? You had to create the story, create the messaging, solve different problems on a daily basis. So I really enjoyed that. And that's when I said, okay, let me start talking to some friends who had some small businesses. And Jason Edelman, who had worked for me, my previous employer, he was leading our networking practice. And he had left in 2014 to start Network to Code. And I stayed in touch with Jason in 2017 was when I was having conversations with him about the company. And there were a few things that just really, really drew me to Network to Code. The first thing was… was that Jason would constantly refer to community. And community was new to me at the time, right? Everything had been proprietary and… But I started just searching and taking a look. and seeing what was out there. And I could really see this group of passionate folks out there who really wanted to transform the way networks were operated and managed by leveraging software and leveraging programmability and automation and leveraging open source technologies. And I could just see this momentum building and this– a group of passionate individuals just like evangelizing and preaching it and that was growing. So that really intrigued me. The next thing. that drew me in was when Jason was talking about his customer list. They were a fortune 500 global organizations and Network to Code was five people at the time. So that just blew my mind. How was Network to Code? you know, finding these massive companies as customers. And he said, well, that's all through the community. through the blog posts that were written through the open source software that was contributed through the industry events, community events, meetups, a lot of grassroots work, and it was all about the community. So I was excited, and I ended up joining in January of 2018. Jason wanted to continue to focus on the vision, the technology, our customers. and the community. And I had this experience working at a startup company where I was able to work with tremendous leaders in all different aspects of the business. So I knew what needed to get done to grow the business to take it to the next level. Didn't know how to do it, but that's part of the challenge. I would go out and figure it out and find out that I needed. So that took me to Network to Code in 2018. And right away, I said the first– first opportunity we have the company had primarily. in its early days, Network to Code. develop the first five-day network programmability and automation bootcamp. So training was a big piece of our business. These large enterprise organizations were looking to Network to Code to help educate them and give them some of the foundational skills required for network automation and NetDevOps. And so business was largely around training and enablement as well as providing services. to maybe automate a certain workflow or in their environment or work with Ansible and develop some modules for newer technologies or develop playbooks for certain workflows. But there were more one-off engagements. –

Yeah, you know, I do remember Jason reaching out to me like years ago wanting to like… do some training over at ONUG. I'm not sure why we couldn't connect those dots together. We should take a look at it if that's something that… you still, you guys are still doing, you know, it's like, this is a very open community, you know, and that could be really beneficial. But let's focus a little bit on, so you mentioned a little bit about customers, investors, has Network to Code, has that been kind of customer grown or have you guys raised money to fund the company? –

Yeah, we were bootstrapped early on for our first seven years, then we ended up partnering with some investors in Guidepost Growth Equity who are just tremendous, tremendous partners in the business. And the idea was to really help grow the business, continue to invest in. product development R&D. We launched in 2021 our open source network source of truth platform and automation platform called Nautobot. So we wanted to ensure that we put further investment into the platform. We wanted to invest in go-to-market. So that partnership has been great for us. –

Okay, awesome. All right, very good. Then kind of like main problem that you're solving, you know, and I think at a very high level, it's really around network operations and it's how do we automate. a lot of the tedious toil. that is bought both part of network operations and security operations. That seems to be the main focus where I think you guys focus with the capital N network and small s sec ops.

Yeah, it's been a challenge for many years and the challenge is just increasing as networks continue to become more and more complex. whether it be from cloud or mergers and acquisitions and the human element where operations has been so manual to date, it's really presented challenges where human errors and inconsistency in configuration have created so much… some many challenges in managing and maintaining that network.

The whole configuration changed management. It was like the biggest like source of… either security exposures, threats that accompany or outages, just pure outages. So yeah, so fixing those are huge. And I think what the stats are, but they're like really huge. So like 80% something like that, you know? Yeah, a few hours. Are those problems. So that's a great area to focus in. So if we think about Nautobot, in particular– Like, you know, what are the three, I'm not sure if there are three features or three major. attributes to that platform that kind of enable companies to automate more. So maybe why don't we talk a little bit about the business model, right? So there's some open source, there's some community. Obviously you guys, you generate revenue as well. So how does like a JP Morgan Chase or a Bank of America, it's like– any of the large financials that are really close to the ONUG community. you know, how would they consume? you know, Network to Code and how they would value that. Okay. –

That's a great question. So first and foremost, we are a data-driven network automation solutions provider. We deliver professional services and managed services to deliver these scalable network automation platforms and solutions for our customers. Since we were rooted, as I mentioned, in community, Open Source has a… close place in our hearts. And open source is just key to our business. When we launched Nautobot, we said we're going to ensure that this is an open source platform that we're going to make available to the entire community. And our goal was to contribute back to the community as the community has served us since really we started. And we really believe that community is what sparks innovation and change. So we wanted to hold true to our roots and open source with Nautobot. Nautobot to answer your question as far as like what are some of the attributes or features that these large enterprise organizations are looking to. We view Nautobot really as both a network source of truth, which is a critical component in any network automation platform, as well as an automation platform that… integrates with other enterprise systems and technologies to automate workflows in our customers' environments. So first, on the source of truth side of the house, the idea is to have a system that can model your network data and serve as inventory and by leveraging that data. you can ultimately drive your intended network state through automation. But the first step is to ensure that you have a place to document your network and document that data. and then provide that data for automation purposes. on the automation side of the house. Some of the key features I think, you know, what we've seen over years are certainly config compliance. Configuration compliance is a challenge as these networks… We see a lot of technical debt in the networks. And so large enterprise organizations are looking for ways to identify configurations that are out of compliance and bring them back into compliance. So that's certainly a key piece. Also, I'll say inventory management. We see that a lot now. So asset management, right? Being able to track down– technology. We had a global bank come to us and say that they had $10 million of unlocated inventory. So they needed help solving that problem. Where is this inventory? While they were continuing to procure additional hardware, even though they had that same hardware in inventory somewhere, but they just couldn't find it.

So do you guys have like a discovery protocol or is it like, how do you kind of discover like all these? –

Yeah, well we would tap into integrate with, we have some ways where we can go out and discover the networks in production, but we will tap into other technologies as well, other tools that will go out and discover the network and then we can import that data into Nautobot. you know, curate that data and model it accordingly. So what's interesting about, you know, Network to Code our solutions is it's not, you know, we don't believe that one tool can solve it all, right? So there are other tools that we will work with. That being said, And the flip side of that is these large organizations are trying to consolidate some tools and some of the tools that they have only are may provide, you know, two or three features that the customer is leveraging. because that's all they really need, yet they have to go out and spend money on these tools that may now be outdated. But we can either build that workflow within Nautobot or integrate with those tools as well. So it gives us options and flexibility.

So do you guys like have maybe kind of plugins for like, like Cisco has the Cisco discovery protocol, you know, and which is great for like Cisco products, you know, but you might have lots of other products, you know, you know, in your network. So, so do you have kind of plugins for each kind of infrastructure vendors?

Yeah, we have, yeah, we have a single source of truth application that allows us to integrate and ensure consistency and data between multiple tools that are available. We can certainly run automation. via Ansible as an example, Ansible, AWS, maybe there are playbooks that are customers that we're building for customers or customers have that we can tie in to Ansible or we can build within Nautobot via our jobs and develop that workflow within Nautobot. So it really depends on the understanding the customer's workflow, which is one of the big challenges that companies have had, right? They're looking at it from a tools perspective. versus really diving into their workflows to understand how they can best automate that workflow. So oftentimes you'll see companies try to fit. a workflow into what the tool provides versus their tool into the workflow that they have within their environment. So we like to take a workflow-centric approach into our engagements.

Okay, so then, so Nautobot the way that I'm kind of like picking up from how you're describing it is that it does a lot of ingesting, you know, to discover the overall network topology and the devices that are there. kind of you build a database associated with that, then there are kind of tools that allow you to kind of look at configuration, change management, or I guess the very different workflows. So tools for those particular workflows. And then you can start to automate that. So it's like, okay, first let's understand what the overall network configuration is and what the– what the devices are in the network. So you got a kind of a snapshot of like the existing plan, you know, of what you have. And then there are tools that you can bring to bear to help automate, you know, and so I think, you know, you know, and some tools might transcend multiple workflows. But so I think that so is that kind of that basically it I can see that as being really useful for network operations engineers, you know, it's like, I finally I have a picture and a view of what my entire network is. Can you guys like simulate? Since you have that data now, can you start to simulate changes? and what the behavior of that system might be with those changes.

That's spot on. We call the workflows applications that we develop. We develop applications for common use cases and we open source them as well. So the common use cases like configuration compliance or circuit management or remediation or provisioning. That's all available within applications that we develop, but the platform also allows our customers to develop custom applications for their specific needs. And if there are some integration points where customers can integrate Nautobot into their other tool sets or applications that they may have in their environment. to really create the single source of truth platform. as far as change management. And we model the configuration of the intended state in our environment so we can do a lot of compare and contrast what is happening within the operational state, how the network is actually configured versus how we… should be configured. And that's kind of where we will do the pre-post change validations to say, okay, if we made the change, are we expecting, are we seeing the results that we expect? If not, we can revert back. And part of that goes to the DevOps mentality that we have, right? This is NetDevOps, our approach to developing network automation. So… That's a key piece to really incorporate version control and testing and deployment, CICD pipelines. So that's kind of all part of our approach to building network automation solutions for our customers.

Yeah, awesome. That's great. I'm super. We've been always– trying to move networking away from kind of CLI and specialized skill sets into more general purpose skill sets. and especially trying to keep up with how fast developers move in the development of new applications. So kind of mapping it into DevOps workflows and DevOps culture. That's probably a big part of why you guys have been so successful.

Absolutely. Think about the innovation that's happened on the application development side of the house over the years. DevOps has been around for 15… 15, 20 years or so. –

About 12. –

Yeah, 12, okay, 12 years. So yeah, but just the innovation, right? Since then, it's been tremendous about, you know, application development, application modernization, cloud containers. And so much technology and innovation has happened because of the agility and because of this new way of operating. And we think the same holds true from the networking side of the house. But the network side of the house, quite frankly, has really lagged in innovation over the years. –

Yeah, we've been really slow because I think there's… There's a whole bunch of reasons in there. And one of them, I think, is– It goes back to what we were talking about very early on. It's like we both went through these training programs to get up to speed on this new technology. I don't think much of that happens anymore. It's like I'm not sure where a lot of new people coming into this industry get trained. They might go CCIE and try to get certifications. That's probably one major way. But I think our industry, the networking industry is very much an older generation industry. And I'm hopeful that the tools that you guys are providing and that others are providing are more suitable to the ways that newer, younger engineers want to run their operations. You know, so I think, you know, I want to kind of move us along. to kind of start thinking about the kind of AI since we started talking a little bit about the change management, almost like a digital twin kind of concept there. So you guys, you're collecting a huge amount of data there. You have these applications. We've been doing a lot of discussions in the Built for Trust podcast around how once AI starts to make its way into the protocol sets and into the products, then we're going to start to see very different kinds of products like AI routers and AI switches and AI firewalls and IPSes and IDSes and so forth and so on. How does this change your… whether it changes your market. your company You know, what you're kind of like, how are you thinking about how AI is going to impact? Network to Code and also you know net ops in general.

Well, great question. First, I'll go back to– just touch on open source and one of the reasons why we are. advocates of leveraging open sources. It gives the customer the flexibility and control that they need. for their environment. And that's a key piece, right? And what we do is, as you're aware, there is a skill set required, right? It's that hybrid. DevOps/Software Engineer and the Networking Engineer. skill sets that are often lacking within some organizations, right? To truly take advantage of… the benefits around network automation and DevOps philosophy or mindset. You need those skill sets. It's still complex to some of the traditional network engineers that are out there, right? And I do believe that all network engineers should have just the foundational knowledge. You don't need to be a developer, but the foundational knowledge around DevOps and some basic scripting, I think that is just very, very beneficial to network engineers. But AI in general is in our mind going to allow for better adoption. of this technology and without compromising. So simplifying the use of network automation without compromising the flexibility, the control that it provides for customers. So I think we're only going to see an accelerated adoption of network automation and NetDevOps in the future with AI. Now, the one– The one aspect in basically the AI maturity model where we are early on in its stages is really making, you know. usability solutions, right? Like allowing users to leverage their tools and data and information that is gathered and collected. since, you know, we. build our solutions around the network source of truth and data, AI will actually have access to that data to solve problems. And then in the future, as all of these networking vendors and really just any IT infrastructure vendor incorporate AI into their platforms, into their technology. The same is going to hold true, right? Like when you look at automation from a holistic standpoint, there's going to be multiple vendors. There's going to be multiple types of devices in there. Yet, the requirements to interact with these components are still going to be there. So that's where network automation can really stitch that all together. –

That could be an interesting market for you guys too, because you're right. Every vendor is going to have their own version of AI. So you're going to have multiple– AIs based upon vendors. So we're going to have a multi-vendor AI problem in the pretty short order. So having a federated AI, maybe that's a role that you guys can play there. So it's like a way in which the AIs can communicate based upon a common source of truth. And I totally buy what you're saying whenever, I know you said it a few times, network source of truth. I think the only true source of truth. within and within compute infrastructure within enterprises really comes from the network. So, so I think that's a really great place, you know, for for you guys to be. So I think that that could be one role. The other thing too, and I'm not sure how much you guys, maybe you've already been thinking about this, is that I'm kind of thinking of three different entities. I'm kind of thinking of kind of AI, and especially its ability to kind of write code. And then there's policy management. So you set business priorities and initiatives. And then you kind of put that into code. So you have policy as code that enforces the policies that you have. And it could be rules, identities, you know, for individuals, access, what kind of resources they access and so forth. So what are those policies? And then you have all the data that you all have. So if there are changes that need to be made to the infrastructure that maybe is because of policy, then you can imagine that that code, those changes could be written by an AI. And then they can leverage your product to– troubleshoot or to basically simulate what does that look like and then get that confirmation that okay that this kind of configuration change was still going to stay within compliance and also going to deliver on the policy that has been required. So you can kind of see how there is a kind of a triangle of systems there that you guys are kind of right in the middle of that kind of data pipeline and configuring that pipeline.

Yeah, absolutely. And just to be able to write– software artifacts, you know, leveraging AI, right, applying that to some… Change management process and. and policy is going to be key, right? And AI is really going to tie a lot of that together. So, you know, first and foremost, you just think that data is the most important aspect, right? Whether we're talking AI, whether we're talking network automation. So, ensuring you have clean, accurate and complete data is a critical first step in anything that we're looking to do in the future.

So if we kind of like take a magic wand and say, okay, all this stuff is really gonna work out. It's like, we're gonna see a world where. we can have automation at a pretty big scale. within networks and we need them because there's like The last time we looked at this kind of job request or job recs in security engineers, about 750… thousand to a million of them worldwide open recs, you know and network engineers aren't too far below that number, you know, so the only way that's gonna get solved it's not gonna be we're not gonna train all these people, you know What we do need though is automation. So maybe like what do you kind of see in the next like three to five years? What does a typical network operation center look like? in the next couple of years, like three years into five years. Yeah, yeah.

Yeah, I think that sometimes it's easier to see what the panacea is and like thinking, AI long-term, you have a self-operating network, right? But it's a lot harder to see, well, how do you get there? And we are a very, very long way away from that in all reality. But in the three to five year mark, I think network operations is going to be a little bit more on operating the network automation platform and leveraging the AI tool sets to gather all the telemetry data and observability and applying that to policies that are defined by the IT organization or security organization. And you'll see Network Operations Center focus a little bit more on the automation platform itself to drive those changes to really drive the incident response and root cause analysis. Tying in all of these tools that are available and tying in some of the new AI technologies that we're going to see at the time to correlate all of that data. So, you know, the shift in mindset is going to be, all right, how do we operate the network from an automation mindset, right? And not like on a device. Okay, where is that issue? Let me log into that device and fix it. It's going to be, what is the platform telling me? What changes do I have to make to the platform? to push out to my network. So I think it's gonna be a mindset shift and you'll certainly see security teams come together with networking teams and they'll have this AI skillset and technology available to them, but you'll see much more unified approach, but they're gonna focus more on a platform-centered approach. view the platform and let the platform address the network.

Okay, so maybe just a couple quick things. I think we're probably getting close to time. Have you collected any data around what does on average and we don't need to do standard deviations or anything like that, but like on average, How many devices a network engineer manages? in a large organization.

I know some organizations that have a network engineer may manage, you know, a thousand devices per engineer and then others, you know, less that may have a hundred thousand network devices. But I would say a couple of hundred network devices that… folks are managing and that's just kind of a ballpark number.

Yeah, like we actually used to collect this data within ONUG, I think our average is about 1200. So 1,200 devices per network engineer. I think in the next three to five years, I'd like to see that increase by order of magnitude. I think that's when we really know. you know, that we are really automating. So I think the thing is that we're not going to replace network engineers, nor do we want to. But also it's like, and we see this– thinking systemically all across various different industries whether you're a doctor or a lawyer, you know, a physician, you know, what have you, is that if you can take someone with a relatively lower skill set and then give them AI tools so that they can now function as like a senior level person, then you've just upped the intelligence of your organization and you've upped the kind of… impact that that individual can have. And that's one way to really close that gap between not having enough people out there to fill out the jobs that all this work needs to get done. So I think that to me, you know, It's like… a relatively smaller growth rate in terms of the number of people we actually need in the industry to do this, to manage the infrastructure. We're still going to need people. We always will. But to really make the folks who are there much more productive. Then I think a way to manage that or one one way to kind of view that is all right Well, how many devices can they manage? You know the one thing we used to compare this to is like the hyperscale companies like you would have a system engineer that might be able to manage like a hundred thousand systems sometimes even more And we were like… God, a network engineer is doing 1200 and you got a system engineer. that's managing a hundred thousand systems plus And that's because the toolset difference was totally… where the tool sets are totally different. You had a lot more automation on the service side of things than you do on the networking side. So anyway, I'm really bullish for what you guys do, John. And it's like you're addressing a big need in the industry and also now with the advent. of AI tools. It's really up to you guys to… Figure out how to integrate that to even make your product even more potent.

Oh, yeah, it's actively being developed as we speak right now So really excited about it. Nick. I really appreciate it and just want one thing before we Before we head off, you really, really appreciate the community that you've fostered and own throughout the years. It's tremendous. You have a tremendous following and from… Oh. customers to vendors, it's just a thriving community. So it's a nice thing to tell.

Well, thank you, John. That means a lot. That has a lot of weight to it. And just to reciprocate, it's like we love Network to Code. You guys always do such a stellar job. you know, at every one of the meetings or the conferences that we put together, like you bring a lot of folks into the community, you always have a great presence, you always bring it in terms of the demos. And you're right in the sweet spot, you know, as well, wherever wants to know. And so I think our pivot, you know, this year to the AI Networking Summit was really… You know, it might not, it may seem now, 2020 hindsight, I was like, oh, of course, I was right, you know, but like, in mid-2023, it wasn't. So anyway, so I want to congratulate my staff. Great. Well, we look forward to New York. Awesome, great. John, thank you so much. A real pleasure. Thanks for this time.

Oh, you're welcome. Thanks for having me.