Networking Field Day Delegate Roundtable: Networking Needs to Evolve

I've, I've said for a long time, networking is one of those weird things where boring is good, right? As long as things are working and it gets a little bit faster with every refresh, nobody really cares. But we have turned a corner in a lot of cases because now the network, more than ever, is very important to the way that we do things.

But is it really, because I feel like a lot of people are saying, we need more speed, we need more capability, but they don't really, in fact, a lot of these folks would probably be way better off if someone would just turn that networking over to someone else. And if you watched the podcast that we put out prior to this event, uh, the one that actually came out today, uh, you probably saw our thoughts on outsourcing, but I wanna take this one step further. Networking needs to evolve.

Either you need to let someone else run your net, just good enough network and keep it operational. Or you have to put the effort in and make your network the centerpiece, make it the product that you're selling, not just a delivery mechanism. So I'm gonna open this floor up, and I'm gonna start with Chris, because Chris is the one who is having this conversation with, and you obviously have thoughts around this because you are heavily involved in, uh, auto Con and the Network Automation Forum.

And so you've seen these people try to push this narrative that the way to make networks evolve is to use network automation. So what are your thoughts on this topic? Yeah, well, I think that's accurate.

I think that, and, and this is using network automation as a pretty large blanket term. I think what we're talking about is the evolution of network operations, right? So the way I look at it is we're in this third era of networking.

Um, so the first era was, does this work, right? Can this work? And that was, you know, we're talking about like the 50 sixties, like, we're like inventing P-C-P-I-P and figuring out, and like learning that that was going to be the protocol that actually worked for a worldwide interconnected network, right?

Um, and then we got there, right? And we got to a point where we're like, okay, this, this pretty much works, right? And then the next era was, you know, how can we make it work better?

Which I think was more about user experience. Um, and this is, you know, taking us from, okay, we have internet service that's possible. We have network service that's possible to, okay, now you've got, you know, an expectation for network connectivity everywhere, even, you know, 700 miles an hour or 30,000 feet.

Yeah. Um, which, which was crazy at the beginning of my career, right? And, and is now just like, oh yeah, of course.

And I think now what I mean by this third era is we need to make the operator experience better. Um, right? So we, we got the user experience pretty good, right?

Most people just kind of forget that the network's there, they just expect it to work all the time. It's really weird if it doesn't work, as you were saying. And now it's, we need to kind of take that lens, I think, to the operator experience.

And I think that means automation in some form, right? It means operating networks as a system, it means operating network using software. It means, you know, using tools instead of just, you know, hand jamming, configs into individual devices all over the place, all the time, time.

But Let me, let me jump onto that point, because you say that we need to make the operator experience better by adding automation. And I know for a fact through all the panels that I sat on eight, nine years ago, that automation is gonna take my job away and it's gonna make things so much more difficult to work on. And you're making my life a nightmare by telling me I have to do all of this stuff to make the network repeatable just to put me out of a job.

Why would you do that to me, Chris? Well, But so it, it can't be all those things at once, right? Because either you're Doing No, that's what I was told.

It's all of those things at once, right? It is, it is Schrodinger's network. It's both complex and simple, right?

It's both making me do more work and putting me out of a job. I don't own any cats that are going in boxes like that. How does this work?

Yeah. Well, so I mean, I think there is more work to be done, right? And so we're talking about almost every network that I've ever touched, whether, you know, whether it was, you know, as an employee or as a consultant has been under documented or misdocumented, right?

And so, so that's one of the first things that there is a huge amount of work out there I Think people would kill for under documentation. Yeah. Right?

I'm think documentation undocumented because these tools, like what we just saw from a visa this morning, like, these tools don't work without the data, right? Um, and so there's, there's a, there's a, I don't think this is like eliminating any jobs. I think it's potentially over time making the job more valuable, right?

Uh, making the job easier. And I think the, the decision here isn't necessarily for, for an organization, at the organization level. It's not do we need automation?

It's, is the network a differentiator in our business enough that we need to build automation capabilities in house? Or is my network truly just a commodity, just a utility? And I should let someone else focus on that who's really good at it?

And I think that is part of the core conversation that we need to be having, is there, we have to make a decision on this. How important is my network? And I'm gonna say this to all of my wonderful friends out there.

Turn up your headphones. How about your not wonderful friends? Well, those too, uh, because I'm about to make some of those.

Do not listen to salespeople when they tell you how valuable your network is, because they're going to try to convince you that it's just valuable enough for them to be able to make that vote payment next month. And you know what? I realize that you only have 10 users in your office, but you really need this wireless controller system with wifi seven aps that broadcast in every possible, uh, radio range.

Because your two room office needs the best possible automated connectivity. And I hope a lot of you out there are shaking your heads and saying, why, why? Because we have convinced people that they need that.

And that's the way that tech goes, right? Anybody who's ever waited with bated breath for the next iPhone or, uh, Samsung Galaxy phone to be released with a whole bunch of features that you're never going to use, yes, I have to buy the pro model every year because it has the best titanium coated, uh, sapphire crystal lens programmatic camera that I think maybe three pictures a week with. But, but, but it's the race, right?

I have to have the biggest and the best. And, and we're not just doing this from a sales cycle. We're doing this with supported case studies.

Anybody in here, remember that time when everybody kept talking about how Netflix was a tech company? They're, they're a tech company, not a media and entertainment empire that delivers media to you. They're differentiating on their network.

They're talking how they're leveraging CDNs for delivery and things like that. And I don't hear a lot of conversation today. Well, I think that's the key point though.

I think it's, it's not important versus unimportant. I think that the network is fundamental. Like, like none of this stuff works without the network.

We network isn't important. We're you can get away without having one. It is it different, right?

It is. Does the network make your product or service different? Um, not even necessarily better, but it, it, does it have an impact on what you selling to your customers, um, in a way that you need to differentiate your network and build it differently?

And if not, then it's not that network engineers jobs are gonna go away, but they may shift, right? More towards service providers, MSPs, that kind of thing. Anyone else?

Thoughts on this? I mean, The network is, I think, fundamental in the same way that electricity is fundamental and the ultimate, uh, metric by which the power grid is, is defined as uptime, right? Network is uptime.

That's the differentiation is whether or not it works constantly and all the time. And there's never any issues. Um, where I see with where you're talking about going with, uh, you know, a Nile or whoever, is it, is this a differentiator?

Is can you live with an hour of downtime every couple of months? Like, I don't know, some companies, yeah, that's no problem. Other companies absolutely not.

And to quickly define for everybody at home, when he mentioned Nile, he's talking about a class of providers that are offering something called network as a service. We've talked to a couple of them here at Tech Field Day before Nile Secure and, uh, meter. There's also a third Ramen Ramen.

We'd love to see you at a future field day event. Um, but the idea is, is that they come in and they actually deploy all the equipment for you in, in the same vein as a managed service provider, and they handle the network for you. Yeah.

And, and I don't wanna pick on them and say that they're, they're bad or they're gonna give you downtime or anything like that. I don't, I don't mean that. Um, but what I do mean is, is that, uh, there is a, you're paying for a level of service there, um, and the more you outsource, the less you control that level of uptime.

But do, are you guaranteed to have that much uptime if you don't use them? No, but there is, you know, uh, um, there's definitely diminishing returns, obviously, but you know, in general, the more you pay, the more you, the more experts that you have running your network, the more you can kind of, I won't say guarantee, but uh, hopefully improve your uptime, right? There's a lot of companies that spend a lot of money.

That True not, I'm not saying every one of those nines in the decimal point triples the cost of your network. Sure, yeah. And, and, and past a certain point, there's a reason why nobody quotes like seven nines is because that's literally, it's always up.

Absolutely. But that even there, you have that conversation of, well, why does my network need to be up all the time? Netflix Netflix's does?

Because Netflix delivers content 24 7, the Whataburger network, they could probably go without it for like an hour in the, like from two to three when nobody's buying the food, right? Even though they're open 24 hours. Like there's programmatic downtime, like, like Chris brought up, like, I don't need to be connected to the world 24 7 if the wifi on the plane doesn't work.

Am I annoyed? Yeah. But it's not the end of the world happy.

But if, but it let, but let's, let's take that step further. What if the plane was suddenly like streaming telemetry? Because that's something we've talked about a lot, right?

Is why, how could you lose a plane over the ocean? Shouldn't it be telling people where it's at? Planes don't stream telemetry.

They, they actually data dump when they land. But if suddenly that streaming telemetry and it doesn't have a hundred percent uptime, somebody's gonna get mad. Yeah.

But you know, it's like, I, I, I think about in healthcare, you know what a CIO that I worked for you said, Hey, you know, it's never gonna stop that interaction with that patient in front of you. You're still gonna give the care, even if the systems are down, you're still gonna give that care. You know, so that, that five nine saying, you know, it kind of goes out the door a little bit, but at the same time, that same doctor's screaming and yelling at you, telling you that Yeah.

That the system's down. So in terms of like, we talk about, you know, commodity too, and we talk about uptime. Isn't uptime just part of that, you know, level of service.

So more back to kind of the point, you know, does this all have to be kind of homogenized and delivered as a product, you know, as a service versus customized for that business to, to exist? But a Lot of it is already homogenized, right? How many of you guys are running ARC now at home Token ring Fitty, please tell me somebody's got a fitty ring.

Thank you, Dan. We, we've homogenized this. Like when I started my networking career many years ago, um, I was still installing serial lines, right?

Like, I still had to go out and install T ones and DS threes, and now we don't do that anymore. I would argue that SD WAN only exists because we standardized on ethernet SCP. There's no way that a velo box or a Vitello box would've worked if we would've had to, you know, clock source line the card in it.

But we, but by mo Chris, it'd at least be more expensive. That's true. It would be a lot more expensive.

But as we moved to that point where the, the components became a lot more homogenous, like, are you gonna tell the person, okay, the blue cable needs to go into the one marked cereal and the orange cable needs to go into the one marked ethernet and they're gonna go, but they're the same cable, right? Yeah. Like we, we've, we've made it easier to do those things, but we've just moved where the complexity lives in the network.

Because now if everything is ethernet, there's other challenges to solve, and you use A straight through or crossover, You Don't ask that question anymore. No. Right, Right.

Because we just assume that the switch is smart enough to understand what it needs to cross. Yeah. And I can still remember walking into the bowels of the IBM plant in Rochester, Minnesota, trying to find a crossover cable to wire a switch underneath my desk.

And there was like, it was a speakeasy, like there was a door that slides. I was like, what do you want crossover cable? Who sent you?

Like, that's the thing is, Must be orange. We, we do these things because we wanna make the network survivable. We talk about uptime, we talk about reliability for our clients.

Healthcare is a very interesting example when you think about it. 'cause the doctors, even though they kind of work for the hospital, they don't really work for the hospitals. They're independent contractors in a way.

And they're your client base that are really mad when the network doesn't work. And I, you, you would say yes, patient care matters. I will say that a lot of nurses that I know, if their little rolling cart doesn't operate, the level of care is impacted, maybe they're still gonna care for that person.

But look, look at the fact that you've gotta scan the bracelet. You've gotta scan the medication. You've gotta double check the dose.

You have to use the smart card to verify that you were the person that actually gave this dosing. Why legal mandates something happens, you get the wrong dose of medication. They have to be able to audit that trail all the way back to figure out what screwed up.

And so a lot of this is being mandated to us through those methods. I mean, it's a spectrum, right? I mean, fundamentally, I, I think that there's no easy answer because it depends on the business you're in.

Depends on the criticality of, of communications. Depends on how digitized you are and how dependent on that digital infrastructure you are. And, and, and based on that as a business, you decide, you know, where you need to run your network.

And at the same time, on the sup's, that's the demand side. On the supply side, you have, you know, your nis, your meters, your ramen coming in. And they're not new.

'cause we had managed service providers before this, you know, the communication service providers, the at and t and Verizon would sell you managed network services. So Yeah, but how good were they? Yeah, it's not, it's not bt I mean, it's not new, right?

So you had that already, but then as a business owner or as a CIO, you're making a decision given the business you're in, given the unique nature of your communications infrastructure, given your cost and the ability to recruit and retain what's the right answer at this point in time through the evolution? Because both are evolving at the same time, right? Your demand needs are changing.

Supply needs are changing. And now when something goes down, we don't worry. My, my wife doesn't care if my wifi goes down, picks the phone and turns on tethering and then everything's up.

Like she has a backup plan where she didn't even realize Well, She, she realizes clearly because, you know, it's Like my, my friend Jeff Fry, Jeff shout out to you buddy, um, was talking about the fact that he just found out that T-Mobile offers a $20 a month survivable backup system. And it's, it's only designed to carry a certain amount of data, but it can fail immediately. Right?

Because you have two WAN ports on your, on your home device. And so if it detects loss of signal on WAN one, it can switch to Wan WAN two and you can use it while they get it back up. I would've loved to have had that a couple weeks ago when I had a, uh, a root cause issue.

Literally a tree root caused my internet to go down for like two days. Wow. But like, the thing is, we, we just expect that, right?

Like talking about, um, you know, the building a network for, uh, a, a, a connected organization. Think about the number of times that you've gone to an, uh, a food eatery, like a pop-up eatery, and they have the all-in-one square register. It's, it's molded plastic with a touchscreen on it.

And all you gotta do is plug that thing in. Mm-hmm. Either ethernet or connected to the wifi and it takes care of the rest of the connectivity back to square's payment methodology.

And, and they don't need to worry about that Many cases. They have a backup, right? Yeah.

They'll, they'll go over LTE if, if needed. And in some cases, if you need it, they'll have two LT 5G LTE links. And so if one providers down, the other one's up depending on criticality, right?

Yeah. So the devices are pretty smart now. Oh, But a Aaron has one real quick.

Yeah, well I talked to Jeremy Phillip at this year and um, it's pretty interesting. I'm in healthcare as well and he is not right. He had been Hyde in finance that we're talking about, um, you know, nanoseconds are too long for his transaction, whereas, you know, I don't know, six, seven day probably fine for buy side.

Obviously he can never anything, uh, to somebody will. Right? But I can, sure.

Oh, Hold on Aaron, I'm gonna ask why can he not outsource anything? 'cause Jeremy, um, one second of downtime million, then million and million million the dollar transaction on. So no way To get an SLA that good.

Just write an SLA that says that I can you down. Somebody will write that. SLA you will not like what it costs, but somebody will write that SLA, But somebody will pin for it.

No one will ever reach it because there's no one to yell. You can't get at your own employees, right? Ooh, hell a writer.

But there's another layer of, oh, you don't eat that, right? That's so in there, or, or here's real quadruple your polyp because you added that later. Stick a pin in that 'cause we're gonna come back to it in just a second.

I wanted Mike to be able to get something in here before I switch the subject a little bit. Yeah, so the, this, I was just gonna echo this basically. And that, you know, same, same CIO said to me, Hey, nobody's gonna care about your stuff as much as you're gonna care about your stuff.

And I think that's the fundamental problem that we're hearing here, is that nobody has enough trust in a third party to take over that responsibility. Because they know that even if they write the LSA, they probably have no ability to act actually execute on it. Even you wouldn't have that ability to execute on it.

But you're gonna respond a lot quicker. 'cause it's your stuff. You Have a critical network, right?

I mean, not for your home network, not for the vast majority of the networks out there, which someone probably could, will run it better than you can, uh, cheaper than you can potentially. So I'm gonna, I'm gonna say something extremely unpopular to the most of the community out there to the CTO of your company. There is zero difference between you doing the network and somebody else doing the network because to the CTO, you are both third parties.

The difference is where you get your paycheck from because to the CEO, the CTO, who answers to the CEO, you're both sit telling somebody else to do something. The only difference in their mind is, can I yell at this person for not doing it right? And does this person owe me money if they screw something up?

Most of the people who are actually totally okay with the network crashing, as long as they get money or compensation of some sort, at the end of the day, they would love to outsource because maybe they don't like yelling. But there are people, and I have been on the receiving end of this on a number of times in my career, where if you crater the network or delete a mailbox or something like that, it's not important whether or not it gets recovered. It's that, you know, you screwed up and I got to yell at you for it.

And, and how many times have we heard? 'cause I, you know, now you work for a large var uh, the throat to choke model, right? Hire me because I can be the single throat to choke when something goes wrong.

By the way, that is a horrible way to put it. Don't ever do that. But like, the thing is, people want to blame someone when that things don't work.

I'm guilty of it too. As soon as my internet went out, the first thing I did was run all of my checks and now I know where the problem is. So the first thing I do is call the provider.

And now I'm mad because the provider doesn't immediately know everything that I know. And they took two days to get a truck out there because it, the, it went down on a Saturday morning and they couldn't get anybody out until Tuesday morning. But that w would it have been better if I had been running my own cable to my backyard and then doing my own tree trimming?

Or would it just been a lot of hassle for me to prep for something that may never actually happen? And I think that that's the trap we're falling into. 'cause Roy's point is exactly right.

Most people who mess around with their networks don't need to. Yeah. Anybody who's ever complained about why, why does the running shoe that gets released this year have to have a slightly different foam in a different colorway?

Why does the foam that gets released this year have a slightly different color on it and a different charging port? Because people love to change things. People feel empowered when they get to monkey with stuff.

The buttons on the sidewalks in the average city that allow you to hit the crosswalk button, most of them don't actually do anything. They're there for the comfort of people that wanna push buttons, feel in control. I I, I want to feel like I'm in control, Dan.

Oh, I Was just gonna, I like pushing buttons, but since, since I'm talking, I'm gonna, I'm gonna throw something out. Um, one of the things that I've seen a lot as we talk about trying to commoditize this stuff is a lot of organizations view it as a cost center. Mm-hmm.

So like, if we start talking about network as a service, I, I've been in environments. We built a warehouse. They had a multimillion dollar budget for the warehouse.

And then they're like, Hey, you need to put switches there that comes out of it's budget. We didn't know there was a warehouse coming. We don't have budget for switches.

Your warehouse isn't getting switches. They're like, no, you need to provide them. Well now they're mad 'cause we went over budget.

So maybe as part of your warehouse, you are consuming resources from a network team. Those resources are gonna cost you 10 grand a month. Now all of a sudden being able to make that conversation of in-house or outsource, if it's gonna cost me 10 grand a month to light up a warehouse, is it, do I do that internal?

Do I do do it external? It's being able to make that, essentially it's a cost analysis. And then like you said, who do you, who do you yell at?

But if you just look at it as, we've got a network team, they've got a budget, I'm gonna yell at them when they don't deliver and hold them to the budget that they can't meet because they don't know what we're doing. Um, like it, you're almost put in a position where you can't effectively figure out what that is, what the, what the cost is, what the benefit is. I ironically, that idea of the internal chargeback for services, I think a lot of companies have got used to it with employees and saying, yeah, you are gonna pay however much a year, 800 bucks a year per employee to the IT department to cover the software and the services, the mail server and all, you know, the office 3, 6 5 and stuff.

I think that's being accepted at the user level, but at the network level it's not. And then when you try and do that, it's difficult because we are not a single point source commodity. Like a user gets a laptop, a, a license for the software and so on.

Were shared use. And people are, you get into an argument about, well, I don't use as much of the network capacity as they do. You get one 48th of a switch.

I mean we, we, we've had similar things in the cloud, right? You try and figure out how do I bill for direct connect? How can I do internal chargeback for that?

And people want to know how much data their v VPC sent over the direct connect in order to propor apportion that bill back. If you're gonna try and do chargeback, and you're right. Network therefore is seen as a cost center 'cause it's easier to just go, just pay for it.

Jody, what was that point? I was just saying take away that access and see how important is to the Well, and, and there's, that's, that's the reason I often tell people that, that the network team we're kind of like the janitors who clean the restrooms, right? When we're doing our job and everything's working great, no one thinks about you and no one knows your name.

When they cut back the funding given to that team, uh, you start cleaning the restrooms may maybe three times a week and then two times a week, and then people start getting really peed off because suddenly, oh, peed on. There's no, there's no, there's no toilet paper. Ironic, right?

There's no toilet paper or the floor's dirty. And then they're wondering why. And it's like, well, because you didn't want to pay for the cost center.

It's actually a quote from a video game that stuck with me for years. If you've ever played the old Republic, there's a great scene where one of the intelligence heads is like, we're sanitation workers. We clean up after the users and the executives, and if we do our jobs right, no one will ever know.

Right? And that's the thing. But when you think about this, compare networking to the other utilities that we, we use, and I have god bless, I work with scouts and I ask them what kind of utilities they have in their home and they're like, I, I don't know, uh, air conditioning, how does the air conditioning work?

It blows cold air. You're looking for electricity. Electricity and water are your utilities, right?

We don't worry about electricity in a building. We don't worry about water in a building. Like if I go out and I decide I wanna buy an electric car, I don't worry about the, the cost of it other than the installation cost of putting the charger in my home as long as my panel will support that extra voltage, right?

Because most houses are over-provisioned a little bit because you can consume additional re resource, right? Like everybody who lives south of the Mason Dixon line knows that my electricity consumption is gonna skyrocket from about May through about September because you've gotta run, you, you can't live without air conditioning. The system doesn't care.

We don't have to worry about repro provisioning a warehouse and buying a whole bunch of new equipment and doing all this other stuff. Water is the same way water consumption goes up or down. But I don't need to worry about repiping the house when I want, you know, unless I wanna install a new sink or something like that.

If suddenly my kid needs to take two showers a day instead of one. I don't need to completely renovate my kitchen to make that happen. But yet in networking when someone's like, we are consuming a little bit more bandwidth because we installed this AI agent over here, I'm gonna need you to go out and spend $40,000 on brand new switches so that it works a little bit faster.

No other utility has to deal with that. Yeah. But you don't have a shower head to use this a hundred times more water because it cleans you better, right?

I mean, Because Nvidia sold me a shower head that suddenly is like, it's dumping the equivalent of a hurricane on me at once. Exactly. I mean, it's the opposite in fact, with conservation, right?

I mean, you actually use less water in less electric city. I mean, because, you know, 60 ward bulbs have now in LEDs, right? And so aside from a drought rolling back blackouts, you don't usually care about utilities.

But on the other hand, a networking, I mean, look at a speed that we've gone from, you know, 10 megabits to a hundred to now we, we think a gig, no, two and a half gig, right? M gigs, you know, 10 gig to the desktop. Of course we need that.

I dunno why. Actually we don't. And that's the thing We don't Yeah, The campus is pretty much leveled off.

Like, unless you buy like a very specialized workstation to do like video edit graphic stuff or vr, you get a gig to the desktop. It's the data center where all this stuff has exploded. That's right.

And, and you know, 800 gigabit ports. Yeah man, we're we got those. Uh, but it's because the workloads have changed, right?

Yeah. And I think that kind of to your point, we are at this inflection point, not because networks aren't working, it's because the things that are utilizing those networks at the high end have changed the way that we do networking. Sure.

And I think, you know, going back to that idea that, you know, the network being fundamental to all of this other stuff and, and whatever's gonna happen, right? I mean, um, I've heard CIOs talking about, you know, having a one-to-one ratio of agents to employees at some point in the not very distant future. And at that point, I mean, this is a little bit off off topic, but what, what is the role of the CIO in that world when half of your employees are actually in the machine, right?

Oh, he's gonna be script, right? And, and there was a company that recently just combined HR and it, right? We've seen that already.

But, but anyway, so, so outside of that, um, I think the, the demands on the network continue to evolve. And that's one thing, right? So to, to Royce point and yours, right?

The, the water demand of the average building and the average house, the power demand of the average building and the average house aren't going up, but bandwidth demands and latency demands continue to, to scale. Maybe they won't forever, but they are, and the number of devices continues to scale. So, and, and I think because of that, that's why, that's why we need automation to make the stuff more, um, usable with the same amount of people, right?

Because we can't continue to scale the size of the network team algorithmically along with the demands of the network, right? And so you've gotta do something with automation, which is how, that's the only reason we can have as many VMs as we have is because we figured out how to use software to run these things, right? I mean, the, the, the way a system administrator ran physical servers in a, you know, in a rack in the closet is very different than the way a system administrator runs VMs on AWS today.

That's true. And we're, or I think we're going through that transition of networking now a little bit later than they did. And so something is going to change, right?

This is going to change. I think the question becomes, to me anyway, is are you willing to invest what's needed now to be able to operate your network the way it's gonna have to be operated in the future? Or would you rather let someone else do that?

The the answer is going to be, I don't wanna pay anything right now until it starts Ty, right? And this, and that's, that becomes this middle road, which I don't think is the right path, right? I think you need to choose, right?

Are you gonna, are you gonna do this yourself or are you gonna have someone else do it? But it, I think it has to happen. And if you don't, you're gonna run to a wall, right?

So what I had a, I had an example come up kind of similar to this. It was a conversation we're having and there, there's a new retail branches that are getting ready to go online. And the the point of contact says, listen, I don't, I don't want to even think about the network.

I don't want to touch a network, I don't wanna see a network. What I want is a monthly bill mm-hmm. For my network.

And one of the solutions that are being offered is, well, the, it's, it's a variation of network as a surface, but it's basically, okay, we'll partner with somebody to come in and deploy all the stuff. You don't own the equipment, but it's gonna be there. You're gonna pay a monthly fee to use this, these networking devices to have connectivity.

Um, and basically it's, it's one of those, okay, you're not investing this much money. It's, it's kind of like the, the cloud mentality where, okay, now you don't have to buy all this stuff because this stuff's in there. You're, you're network's there, but you're not actually, you didn't buy it.

You don't have a a, a service contract with it. You don't have a a support, somebody is managing it. But it, so there, I, I think there's just like, it's one of those where, kind of like what Roy was saying, there isn't like a one size fits all solution.

There's gonna be consumption models where one company doesn't want an IT staff, they want a monthly bill. Another one's like, well, it's, it's important enough to me that I need somebody where I don't wanna pick up a phone and say, Hey, I need you to fix this. I want to tap a shoulder and say, look, this is what's going on.

Run down there and check on it. Yeah. I, I'm seeing, I mean, physicians group law offices, you know, accounting offices and all that, even they're going the, the NASH route, right?

Let me just pay someone mm-hmm. Because we've commoditized networking in a sense, right? I I give you, you know, a wifi 6 67, you know, I don't wanna worry about that.

Someone takes care of that for me. I don't wanna call up, you know, Cisco or Juniper or you know, HP Aruba networking, whatever, right? I just want a problem to go away.

And I think we're at a level where generally speaking, for most use cases, the network stays up, you know, on its own managed by whatever console, you know, management system. And it's reasonably good and someone else can take care of that problem for me, because I don't really want to own aps or own switches, someone else would do that for me. Then there are the critical businesses where you really need a network special.

You really need a network up and, or you wanna run it in a special way and no one's gonna come in and for the money you're willing to pay, say three network engineers give you the SLA that you actually want. Right? And so you hire them.

Right. Well, and I think that that kinda goes to what we were talking about on the, the, the podcast kind of ties in with that, which is, there's, there, there's an inflection point in terms of size and scale, right? There's a inflection point of saying, uh, you know, you're a law firm with like 20 or 30 offices.

That's something that's very, um, you know, they don't wanna deal with it, right? That's not their core competency. It doesn't matter to them.

It's a consum, it's a consumption model. They want to have that service to say, Hey, we have this here. Um, you know, we get a monthly bill.

Then you go to, then you go down to your enterprise level of, you know, other shops that are selling consulting services, for example. Um, they have compliance requirements. They have other supply demand, uh, demand and supply needs that they need to have that specialization, even if it is just consuming some of the vendor services that, that are prepackaged, but they're still main maintaining that in-house.

Jody Had his hand up, Jody. Yeah. We hope that you kind of raised the point that when he said earlier that the campus network is largely flattened out.

Everything's gone high end in the data center in the cloud. It also requires a little bit of rethinking of to what do we mean when we say the network, right? Because, so the network, the network from the user perspective is anything that point A to point B, but there are multiple networks mm-hmm.

In the middle, all of which have different requirements. We're seeing a lot more automation requirements in data centers and cloud. But on the other hand of the campus network, we're seeing a lot more simplification.

I've had two customers in the last six months look at me and go, our network needs to be a Starbucks with your barista. And essentially all they're doing there is they're riffing out anything that it conflict. They have a primary internet, they have secondary internet provider, they have fast failover commodity equipment.

It's cheap. There's no automation, no requirements for automation. And they move all of that up to the cloud and the data That's actually automation.

The, the cloud management is the automation. Yeah. Right?

I mean, your Meraki, your, your miss, whatever it is, there is automation, it's built into the software. The cloud manager is the automation. Yeah.

But, but they're not doing Anything. No, they, they're not doing anything special. No.

Correct. Because There's no requirement for it. They just keep it simple, keep it dumb, and keep all of the security between the endpoint and the data center or the endpoint in the cloud and shrink what we consider the network to be.

So one of the things that this keeps kind of popping up in my head was, eh, give or take, four or five years ago now, there was a whole bunch of talk of like open networking and SDN and white box switching and basically everything. He's laughing over there. Yeah.

Everything that kind of, you know, like what VMware did with, with workloads, like we're gonna commoditize the server hardware, which we can debate whether or not nobody's going and building super micros for every, like there's still Cisco and all these companies are still building servers, so we haven't quite got there. But basically there was this big idea of SDN was gonna pave the way for, we put in commodity switches and then we just push whatever services down from a cloud managed platform. But it's, it's basically gotten to a point where, well, yeah, we've got me Rocky, or we've got Mist, or we've got insert name of vendor here, which now doesn't like, we've, we've basically built SDN in a very proprietary way.

And kind of like what you were saying earlier with the, the standards based thing. Okay, if I, if I go with vendor A, even though it's SDN and it's their standards, it's proprietary. Yeah.

So my ability to, Hey, I want to go out of this. I've gotta either work with that vendor or now I've got a separate vendor with their own management. Now I've got, it's, we're adding complexity because there isn't that, that kind of simple unified, here's the open standard for how all of these things play together.

We're gonna go white box networking someday. Well, that kind of is a standard. It's just, it's the lowest common denominator between all the vendors, right?

That here. Yeah. There you go.

Now, I mean this, this was the big open config problem. It's like, hey, we can have one kind of lingua franca that, that unifies all of these configurations and, and then all your software can talk that and open config can push it out. It's like, that's wonderful.

But this one has a feature that's implemented a different way than this one. This one has a feature which the other one doesn't do. How do you deal with that?

And so now we end up back where we are with S and MP where everything's under, under the enterprise org rather than under the standard MIBs because, well, everything's kind of special. 3. Which one do you support?

Well, they're both open flow. Oh, but they're differences. Which one do you program is probably the more important Question you shouldn't be programming, I think, I think what we realized with SDN has been there, the whole journey from the very beginning to be, it never ended.

It just got sub steam in the other architectures. I think Dan, as you pointed out, it is there. Well, And to be fair, that was kind of what the same with OpenFlow, kind of what we told the vendors at the time when they presented, Hey, this is wonderful.

It does this. We said, but the hell wants to program that. I don't wanna have to write the logic to figure out how to push Mac addresses around on a switch by programming OpenFlow.

I, I need someone to do that. And so I need that as a service. Agreed.

And so on the campus networks, I think as, as Tom was saying on the campus networks, the SDN principles and all that all built in, but we realized in fact that the openness of it, or the white box nature of it wasn't really the problem you're trying to solve. The problem is, I just want a cheap enough, good enough campus networking solutions that stays out of the way, so I don't even worry about it. Right?

I think that's what we wanted. And that's how Meraki and missed, and HP, Aruba and all these guys came along and said, oh, okay, does it have to be white box? Customer says, I don't care.

Right? Just make it look. And I think that you, you bring up a good point that I kind of want to turn this conversation into, get back to the original point about that middle path, because I have an idea and it, it actually involves the matrix.

So work with me here. I'm scared. When you think of a modern network, just in your head, when I said that word, what did you conceptualize?

Was it any, look, anything like an enterprise? What does an enterprise look like? We did a podcast about this.

I'll tell you that most of those networks look like a dentist office in a law office and they look like a restaurant. Those networks don't grow because they don't need to. They're, maybe we added a new server, maybe we added some credit card tip machines.

They don't grow. They are the perfect targets for this network as a service model. This build a utility and let it run model.

And that's why companies like Meraki went after them. But what about those other companies? You know, the ones when the CEO goes to a trade show and then comes back to your office and walks downstairs and said, this guy showed me this really cool thing and I wanna be able to do it in six months.

And you go, we're not ready for that. Well, I ordered an Nvidia Haswell cluster, it'll be here next month and I need you to figure out how to make it work in the Matrix. They talk about the fact that not everybody accepted that they were plugged into the matrix.

And so the way that they weeded out the people that would crash the system was to give them the illusion of choice. The architect and the Oracle figured out that if you let people think they have a choice as to whether or not they're plugged into the matrix or not, they'll choose whether or not there was actually a choice there to begin with. I would posit to you that the reason why the middle ground exists is for those CEOs that think they have a choice to grow this company 10% year over year for infinity, when in fact they can't.

And they won't. Whereas a law office and a doctor's office and a dentist's office are bounded by the number of people who are providing care and the number of people working there, they will never grow 10% year over year for infinity because they are not designed to. But companies that are investing in giant AI clusters, people that are building these robust automation systems, and they're leveraging them to try to open new lines of business, they're leveraging them to try to pivot what the company is doing because this is the shiny new thing over here, and I want to do it because this talking head told me I needed to, if I don't want to get left behind, they're the ones that are currently in the middle road.

Should they be the doctor's office, the dentist's office, and accept the fact that we're gonna stay in business, but we're not gonna grow 10% year over year for infinity? Or do they jump to that? We need to make the network a showpiece part because we have to install all of these new, uh, GPU clusters to build AI agents to consume our data, to provide blah, blah.

You've heard all of the stories and you know exactly where this leads a whole bunch of Nvidia GPUs sitting unused in six months because it turns out they either couldn't make the network work, they couldn't afford to run all this stuff with the new increased power costs. Or it turns out that there's really no end goal in sight other than just patting NVIDIA's bottom line. Do you think that the middle path is merely an illusion of choice for people who are going to make the wrong choice?

I dunno if they're definitely gonna make the wrong choice, but I do think it's somewhat of an illusion. I mean, and, and actually some of the things that have come up here is that we're already going down one of these two paths, whether we know it or not, right? Mm-hmm.

And, and in that, but as Long as the, as long as the middle exists, I have an off ramp. I can make a different choice. Yeah.

Yeah. Even if you can't right. Or shouldn't.

Yeah. Well, it's interesting 'cause I mean, 'cause to the points we've made, right? Like Meraki is, is somewhat of outsourcing your network, right?

I mean, maybe it's not to a bunch of people, it's to a, to a, to a piece of software. Um, which is interesting, right? And then same thing, again, kinda looking at the systems world again, right?

There are still systems engineers needed in every company, even though most companies aren't running their own servers anymore. Mm-hmm. Um, and so I think that's the thing too, is looking at this, how does this evolve?

Like whichever path you choose there, it doesn't mean that you don't have somebody running a network at your company. I think you still absolutely have to have that. I think what they're doing is to Jody's point is they're the ones who's looking at the big picture of, okay, how do I stitch all these networks together?

Um, because I think most modern enterprises, most of the networks they're using, they don't own No. Um, which is a different problem. That that's a very interesting problem, I think.

Um, and so then you're, you're abstracting up a little bit. And so I think in any case, um, there's organizational changes that are needed to recognize how networks work today, which is either, Hey, I'm, I'm building my own auto, I'm, I I want to go as cheap as possible and have as much control as possible. So I'm, I'm, maybe I do go white box and run Sonic on it and build my own orchestration platform.

And I need that because my network's changing and dynamic and I'm dealing with financial multicast or, or you know, subsecond ad uh, selling right in ad tech world. Or there's, there's various reasons why I may need that. Um, and I need to adapt to my organizational culture and structure to be able to operate that system.

Or I'm looking at this, I'm piecing all these things together and whether it's Nile or Meraki or Mist or, or whatever it might be, I've gotta figure out a way to, to build an organization that can manage all of these different networks in a cohesive way. And so I think that's where, that's why I think it's not a, I think there is no middle road at all. Um, you're, you're, it's a false idea.

Like you've gotta go down one of these paths. Are you gonna be managing systems and, and managing outsourcing partners and, and managing the network in that way? Or are you really gonna build this thing, roll up your sleeve and build it yourself?

And to Jody's point, again, that could be, there could be different answers in different parts of your network. I think it's, you you're familiar with wardly maps, right? Yeah.

Like you think about like, that's how I think about this, right? It's just to really think about this, you know, what is it that we're doing sketch map down through Where are the different parts of are, so the network's this low level part of what we do, and is that a on the commodity end of the scale commodity utility or is that some special unique thing that we really need to do, aim to accept that? Like you're just saying there, maybe for different parts of our network, there's different things we should be doing.

Like I think about what we do internally. Say we had traditional, uh, on-premises based wireless controller stuff and it was a massive pain. We've moved it over to one of your, uh, cloudy platforms.

It's fantastic. But for me, the wireless network is not a, it's not an important part of my network. I don't really care about it.

The other bits of the network where I'm doing some quite a little more interesting things, that's much more custom tooling, our stuff, our things. It's all and see, which is great though because like I said, there's other utility bits about it. I don't care.

I don't have to touch them. I don't have to. It just works basically.

Because both can be the same, right? A VAR is a very good example of this. Their internal network can be a commodity, but they can be selling services to somebody else that are a product like a colo facility that they help cooperate or, uh, the service that they built runs on top of their network and they have to do certain things.

It's kind of funny when you think about that. Like you could be a VAR and you're in your, when you're at work, you are network, you don't care. Yeah.

And, but we're the network experts trust us to run, we To kids have no shoes for all the networks that, that we used to put in. Ours was actually one of the worst. Jody has his hand up and then Dan had some thoughts, Jody, go.

So the more we outsource the network and it's not our network, the more we turn back to the matrix and have to face that there. The truth is that there is no network, There's no spoon. Uh, Dan, so Kind of like these guys have been saying, I, one of the things that came to mind was, um, when trying to kind of go through and like teach students what cloud was, and really the, the biggest thing that I always would come back to is kind of the shared services model where when you start looking at all the different pieces, all you're doing is you're shifting who's, who takes an ownership of it.

And when we start looking at networking, like you've got the, the hardware, the software, you've got the, like the, the project side, the operational side. So I've, I've worked for partners. I've worked for, uh, managed service providers where like as an MSP do the day-to-day operational stuff.

But the, the customer owned all the projects. I've worked in environments where we did all the project work and the customer did the, all of the, the engineering, uh, like operational work. I've worked with places that have leased hardware, that have owned hardware that, you know, all of these different thing.

So it's, it ends up being this kind of amalgamation. But the other thing that I, I think is worth bringing up, 'cause we've, we've touched on the electrical grid quite a bit and people mention things like electric cars or AC in the summer, when you're going through and you're building something like that, you've gotta build for whatever that high watermark is. So if you're trying to build out a data center network, hey, I only need one gig today.

Like one of my favorite things was always, well, your peak bandwidth over the last three months was one gig. It's 'cause you only have one gig links. So I'm gonna sell you one gig.

'cause that's the most, like what do you need? Well, hey, there was this one time for half a second that you burst up to 400 gig, we're gonna sell you all 400. Uh, we'll sell you a hundred gig ports and you can be, you know, that, that half a second things are slow.

But it's trying to find out where, where those requirements are, where the needs are, what you're willing to, to let somebody else take the, the responsibility and risk for and what you want to keep for yourself. Right? There's, uh, but there's a scaling problem with all of this.

And it's funny that we talk about the electric grid and I'm thinking, well, like my home internet connection used to be 64 K-I-S-D-N. Now it's one gig symmetrical, right? That's like a 1500 times increase in the past 25 years.

That's quite large. I haven't started using 1500 times more electricity or water in my house over that same period of time. I also haven't had my water and electric supply get end of life by the vendors, right?

And tell me they no longer support my electricity. Your Sponsor No longer works. That's a real problem when you're trying to get any kind of stability in your network is you are forced.

I don't know if if I use the expression painting the fourth bridge, this bridge in the uk and the, the joke is that by the time they've painted from one end, by the time they get to the other end and finally finish the job, they, the start needs to painting again. And it feels like that when we're doing wifi upgrades and infras. Right?

And it's like you never stop. Uh, Nate had something and then Jody has another thing. So Nate, what's your thing?

So I, one more like factor in the equation too is also kind of the, the skills gap that is like continuing to, to kind of grow with the, with the new and upcoming workforce. Uh, we, we have some clients that say that they're, they're new college grads. They don't even want to touch a CLI like for what if it doesn't have an, and not that they wanna click around in a gooey, but if, if I can't, if I can't talk to that thing through an API like what?

Yeah. It, it's legacy. And we should figure out how to, well, We've got prompts for that.

I'm gonna be a prompt engineer when I grow up. Uh, Jody, what was your thought? Well that'll pursuant to the idea that we've got people who are coming out of the university programs and the, like not wanting to touch the LI not wanting to touch x, not wanting to touch y they want a gooey, it's like we got people coming into the industry that aren't accepting that they have to learn.

And this requires that we either shrink things down to the people who do want to learn and do want to do things, have less to deal with, or that the other networks that the people who don't want to learn are, are running, are simple. I don't think this is a good path, but it's kind of where we're, I'm okay if they can do everything via API if that's what they wanna do. It's up to that.

But it requires those APIs to exist too. Yeah. And on that, on that same note, I mean we had just recently our, our automation lead said that the, the skills gap on automation customer, uh, customers that are looking for their, their network engineers to also have automation skillset behind 'em.

He said it's, it's unbearable. Like the fact that, that we're, we're working with customers and trying to implement these solutions and guiding them along this path to where you're gonna have this amazing automation experience, but then the, the people that that they employ aren't, aren't capable of executing on that vision or, Or as Jody said, they don't Want to learn or they don't want. Yeah.

I also, I also think, I feel like, you know, my career there was this fixed window, you know, Size of, of knowledge that I had to carry in the Last five from years. It feels like stuff off the back isn't falling off. I have that old router that's in the corner over there that nobody touches, you know?

And so as these new people come in, they're learning this, this front end edge of it. And there's a bunch of us that are still doing this, but we're also doing all this now. And that's one of the things I think that's going on is that, you know, we're still, is that the people that have been doing this backend of stuff are getting towards the end of the, the, the road and, and they're going, I'm about done here.

And that stuff's still running. And I, I think that's another problem we have coming. I think again at auto con, I I think it was said, you know, that you guys have a diversity problem here that's age.

And I, I think that's the other road we're running into. But The New people haven't figured out that, uh, the APIs that they're using largely just wrappers with the Yes. Yeah.

But I, but I think there's also a degree to which, um, the, the age thing's an interesting side topic, which is like when I, I couldn't hook up a scope to an ethernet signal and tell you if there was a problem at the physical air anymore, but the people who taught me networking could and did. Right? And I just assume ethernet kind of works and if it doesn't, I'll change the cable or plug it into a different port, right?

And the people coming in now, you know, I learned all about IP and IP A and stuff like that. And the people coming in now can't troubleshoot an A problem. Why?

Because it works. 9, 9, 9, 9, 5 nines of the time a just kind of works. Okay.

And so I think everything, when you come in, you kind of ignore all the fluff that happened before you, that the last generation knew intimately. 'cause it just works. And then you move on to the next thing, which is kind of scary.

But that's also why you have to have a blended team really if you've got any old stuff flying around. 'cause the new people never, were never taught that stuff. I think the team structure's a really big part of this.

I think, and to your point as well, that it's the team structure, right? It's this idea that you're probably not going to have these unicorns all over the place that are gonna understand deeply a issues and routing issues and not just, you know, IGPs but also BGP and also understand, you know, the difference between the switch asics and like how to, how to select which, which is I want for which applications. And then also know how to build the software on top of it to run it all with the team of the size you actually have.

Like that's not one person almost ever. And so that's where this organizational structure, organizational culture, like outcomes-based leadership, a lot of this stuff comes into play. 'cause we need to like rebuild, I think the way we're approaching networks, especially in large organizations around these kind of squads that are multidisciplinary, that own a piece of the network and they've got a systems engineer and they've got a programmer and they've got a network person and they can all work together because it just, it's not one person anymore.

I don't think You have to build a network your operations team can run. Yeah. And if you want a different network than what they're capable of operating, you need a new operations team.

Yeah. And, and that's the challenge that people have is because for a long time we've just made it work with what we've got because that's how it has always behaved. And I think that that ultimately is the problem is that this is not something that can be solved with tech.

To Lindsay's point, if you don't know what you need out of your network, how hell are you gonna be able to build it? Does it need, does part of it need to be utility? Does part of it need to be a showpiece?

Until you can answer that question, you better not pick up a checkbook and you better not listen to any salespeople tell you to go buy something because that's how you're gonna end up with a whole bunch of stuff you can't use or don't need a whole bunch of things that your new, um, support and operation staff don't know how to operate and don't care about. And you're gonna be right back where you're at right now sitting in the middle of the road hoping that you're not gonna get hit by a giant bus that says, what are we doing now? SDN ai, uh, quantum whatever, the bus is still coming and you've gotta decide are you gonna get to one side or the other so you don't get hit.

As you can tell, there's a lot of things going on in the networking space, as boring as it is. And we're very thrilled to be able to bring you all of these great discussions and many more at Networking Field Day and all the other events that we do because then the networking space bleeds over into mobility, it bleeds over into AI infrastructure. com and I promise you that there is a discussion going on there that you're gonna wanna pay attention to.

com and up in the menu there's a dropdown that says Delegate. And you can, uh, click the link to say become a tech field day delegate. Fill out some information so that we know who you are and, uh, you might get an email from me wondering if you have time to chat so I can learn a little bit more about who you are.

And you never know, you might find yourself sitting here with one of these fine people debating a topic that we bring up in a round table. We're.