In this second episode of our customer series, we have Avi Freedman, Jezzibell Gilmore, and Joe DePalo from Netskope joining us today to talk about how we as IT professionals, networking professionals, security professionals, systems engineers, cloud engineers, you name it, can take control of our network in twenty twenty five.
Avi, Jezzibell, and Joe, welcome back, and thanks so much for joining me again on podcast.
This is the second episode in our series together. So, I really appreciate you joining me again to continue to flesh out some of these important concepts, important issues that folks in IT and specifically networking are dealing with. And now that we're moving into two thousand twenty five, what I wanted to talk to, about today is how folks in IT more broadly, I'm gonna say. So, you know, those networking engineers and systems engineers, business leaders, folks that are thinking about new tools, new strategies, how to cut costs, how to do operations better in any sphere of operations that they're in, perhaps security as well. I forgot about that one.
And how folks can take control over their IT environment, how folks can take control over their network in twenty twenty five. There's a lot of a lot of things that we can talk about and flesh out here from a technical and a business perspective, a strategy perspective, workflows.
And, and between the three of you, there's a tremendous amount of experience and knowledge that I'm looking forward to hearing today. So thanks so much for joining, and, let's get started. And, Joe, if you wouldn't mind, I'd love to start with you.
Yeah. So, you know, from from my perspective, my background is building wholesale large scale infrastructure where the margins maybe weren't, super rich. And so you had to be responsible for every component, you know, to the hardware, to the layer one, layer two of the network.
But that also gave you control and visibility. And I think as the enterprise network evolves and the market has become more reasonable from a cost perspective, I think the enterprises need to start taking more control, to use your word, of their infrastructure and their environment. And I think through the podcast, we can talk about data, and we could talk about security and other aspects, but I think it starts layer one, layer two, layer three. Right? Where is the traffic going? If you're if you're dumping it off to a transit provider, you're losing control. If you're relying on a public cloud, you're losing control.
If you're paying a managed service, you're overpaying.
And so looking at that, the physical capabilities of your underlying infrastructure is kind of the first step. So that for me, it comes down to not relying on third party vendors to to to to control the destiny of your networks.
Interesting. So, Joe, what you're telling me is that folks can start off, that if they just don't have an understanding where their traffic is going and what that traffic looks like, that's the first step.
Yes. If if you're using a third party provider, if you're using a public cloud, if you're using a managed service and there's a congestion complaint, all you could do is call somebody. Right? You have no control or or capability, visibility. And so, and the reality is people are not provisioning infrastructure for you. So it's just a shared environment anyway. And so decoupling that, having more control of your physical infrastructure, of your routing, of your users, of the destination is is the only way to do it and and to assure that you can deliver a service that your employer, your your users, you know, require.
Yeah. I I would I would, very much agree. You know, it's about observability, which is understanding the assets and what they're doing and the performance, traffic, costs, security. And then part of the key one of the key things elements in terms of controls variety.
So having the ability to have the food to use to be able to, put it all together, because as Joe said, if if if you only have one path or one provider, all you can do is complain to them. So you need to have enough interconnectivity, enough diversity in terms of cloud, and even vendors, routing, although that gets a little trickier for reasons we could talk about some other time.
But, you know, you need to have that, and you need to bring everything together. And we see that as, two big things that people are working on for twenty five is how do I bring all the understanding of what's going on together, and then how do I plan, to have the diversity and and control over our business's destiny.
What what what do you mean exactly by the business's destiny, though? Are you are you talking more about, like, the outcomes of the network as it relates to the business and not on, like, a specific tech my mind always goes to specific technologies, but that's not what you said.
I mean, yes. It's the packets that make up the revenue of your business. If you are delivering your services, if you're delivering support, if you are a SaaS company, if you are a service provider and and, you know, you may be doing higher level things, but you still have to deliver those packets. If you are reliant entirely on third parties to control the quality of them, the security of them, even sometimes outsourcing visibility and debugging, then when you have an issue, your business is stopping and you don't have that control. If someone else has that control, then you don't. If you don't have enough variety of paths and enough understanding about how to use all of those different pieces to be able to route around issues optimized, then, you know, you may be on the hook for something that you know, and then sit there helpless. And that's just not acceptable if your employees can't work together or even worse, in some cases, the customers can't get access to do business with you.
So in other words, we can, we can make technical decisions that ensure, the business outcomes that we want. And that maybe is a is a is a different paradigm for some, technical leaders going into twenty twenty five and getting away from thinking about a specific vendor or a specific technology. Do I need EVPN, or do I need layer two leaf spine? The the well, what are you trying to accomplish, and how does that one or the other support the business in your in your data center in that case? So, let me let me ask you, Joe, because I'd like to get your take on this. I mean, have you seen enterprises in the past, and hopefully now that's the the trend in the future, you know, approach this in that way where they're where they're really taking the business outcomes of the, you know, the whole goal of the network here into consideration when making those technical decisions?
Yeah. It's definitely a growing trend. You know, the the medical and the financial institutions have been running their own networks and data centers for a long time.
And and now, you know, it's not just critical, providers or infrastructure, more and more enterprises. I've seen providers that are deploying and working on their own VPN solutions. I've seen them, completely rip out overlays of MPLS and SD WAN and go, point to point network based services. I've seen I've seen some use dark fiber and and core network services. And so, the one thing about the Internet is being able to have your own bandwidth and your own capacity is going to provide the biggest assurance for you. And so, shameless plug again for Netskope in that, we built our platform as an overlay to interface with enterprises where they need to be, whether it's a public cloud or whether it's peering of BGP or whether it's, you know, PNI of network. But but, you know, to answer your question directly, it is definitely a growing trend of the enterprise operator taking back control of those of those physical layers and, of their destiny of network.
So So can we talk about cost then?
I mean, we're talking about making technical decisions in the context of business outcomes, but what about, cost, which is always a concern? I mean, whether it's a price per port or egress charges from a cloud provider or something like that.
You know, years and years ago, that was that I I didn't hear people talking about it as much, but now it's top of mind, in in really even, very, very small and medium sized businesses.
Yeah. I mean, the the price that the cloud providers, even at deep discount charge for Internet egress, which includes to the other clouds, is more than ten times. It used to be only ten times wholesale transit cost, but it's still pennies per gigabyte, not, you know, per megabit per second. And so that adds up at scale, both for customers that are using things heavily sort of always and if you've got attacks, if you have misconfigurations, if you have things like that. We had a fascinating meeting, about a month ago with an enterprise customer that I thought was gonna be about something else because they wanted to talk about cloud capacity planning, which is awesome because that's a lot of what we do is try to understand, especially intercloud and cloud to hybrid on prem.
But still, you know, cloud, you can use APIs. It's not like you have to order stuff and put it in, so we weren't sure what they meant. And then it turns out that what they wanted to talk about was their own backbone that they built and run and interconnect at, you know, four hundred gigabits per location per cloud in ten cities because, cost and performance, they can't really count on the clouds to have that capacity and be able to run it. And then from there, it looks like, you know, nineteen nineties, two thousands, kind of, like, how do I run a backbone? So it was like about cloud, but it was really about control and running that network. But if you do the math and let's say they've got ten cities times two clouds, or an aggregate, I think, you know, let's say it's, you know, a terabit of traffic. It's actually more than that in this case.
That bill, while it may be less than the rest of their cloud bill, is still, you know, approaching that of if you were to just pay even eighty percent discount with the rate card for that Internet bandwidth is is still, you know, millions of dollars a month just for bandwidth, you know, at way higher rates. So it gets a lot cheaper to bring that under control even if you have two or three or four people that have to be aware of your architecture, and get some assistance in consulting to set that up than to just, you know, pay the clouds for that.
And that wasn't always the case. And so a lot of the enterprise networkers I talked to assume it's too expensive or too hard to manage. And so I think, you know, being able to explore that now is something that they should, definitely be focused on. Because to your point, it's it's a lot more manageable from a cost and implementation perspective.
So should that be top of mind for any, you know, network engineer that's just out there keeping the lights on? I mean, that's that, to me, has never really been part of technical decision making in the sense that, like, I need to know how to configure a thing and how to reduce latency and that sort of things. But, certainly, there's a cost associated with achieving those subs those types of outcomes. Right?
Yeah. But the reality is, like, they're paying for something that that doesn't, you know, the an MPLS overlay or a managed, IP is is not a thing. Right? There's very little benefit.
You're I mean, AT and T has a hundred hundreds of millions of dollars in managed services.
What is that necessary for? If you have one link to them, what are they doing? Right? And so and MPLS tunnels are just, carving up capacity that already exists. And so I'm just very cynical of overpaying networks from an enterprise perspective to give you capabilities that just don't exist. And so, I would much rather go direct to Internet, have multiple Internet providers. Like, Avi said, hire a couple guys that know how to sling some packets, and you're gonna get more capacity, more control, and you're gonna reduce costs significantly.
It's also really important when you go with vendors and manage service providers that you make sure that they're open to giving you the telemetry that you need to make these kind of decisions.
A lot of the managed networking providers, that's all some service that's like Microsoft SQL with a Flash interface that was written twenty years ago that shows you a couple summary points for the day in terms of total traffic usage.
You can get that when you're outsourcing, you know, when you got wireless control planes, SASE, you know, making sure that you can get the data that you need so that your team can bring all that together.
You know, it's enough of a headache bringing together SNMP and streaming telemetry, but, you know, across your networks, bring it together. That's really critical food that, you know, that those teams need as well.
There's something that I actually I was sitting with a customer earlier last week, and we talked about their need, which something Joe and Avi both mentioned. The customers need to have control of their own destiny so that they felt that it was really important to have their own tool that they manage using open source.
But then, you know, as a tool, But then he said, but I also am aware that open source is a gift like a puppy, not a gift like like a puppy.
Yeah.
So As opposed as a gift like a free beer. Right?
Yeah.
Yes. So and and I I think that's a really fantastic thought to leave because when, you know, enterprise customers and even you know, service providers are choosing tools to gain control, they also have to think about the longevity of the and the maintenance of, you know, taking control of their own destiny. Right? What is it?
Do they have the talent in house, the resources, and what's what is it that they need? Do they have the visibility into a thing? You know? Do they have a partner in the vendor that they're working with, the supplier they're working with, or are they, you know, getting themselves into a situation that's just gonna cause themselves pain?
You know, who who is it that you're going to be working with? So and I think that's something I would like for both Joe and Avi to opine on because that's something that's sort of been challenging me because I'm like, yes. You absolutely need to have control over your own, tools and your own infrastructure. But at the same time, right, what's the right way to do this?
For us, it comes down to the people listening to this podcast and the people that are operating these environments are the people that get called when something is broken. And if your vendors, don't have visibility, as Avi said, or some level of control like we've talked about, you know, you're gonna have a bad time. And so as you're you know, if we talk about AI, we talk about security network convergence, we talk about network control.
It all comes down to decoupling the dependency, that each vendor provides you and give you that control versus I just pay them an extra fee to do it for me.
And unless their service is flawless, you're you're gonna have a hard time explaining that to your management or to your users why something is down and why you're waiting for them to update you. Because we can tell you the underlying infrastructure is not, flawless and is not something that could be managed by a single carrier. And so especially if you're a global infrastructure, you know, good lord.
And so that's something that, the network operators have to have to be consider.
Yeah. I think it is important to find vendors that will be open with you.
We've all heard of network marketecture, you know, but, basically okay. Tell me what's behind all the, you know, the dog and pony and the bullshit magic.
And, you know, help me understand and meet and talk with some of your customers, that have had issues and how you know, so I can talk to them about how you've dealt with it because everyone has issues you need to work around. And the question is, are they there so that they're not wasting your time, you know, when you call them and say and they say, oh, there's no known problem with the network at this time. And then if it was a really major outage, maybe the next day, they'll say, that problem we weren't having is resolved now. You know?
Meanwhile, you're trying to figure out what to do. And as Joe said, you're responsible to your management. So it's finding an ecosystem of people that you can rely on with technologies that you understand and finding the most elegant way of combining these things and controlling them. And I am of the bias that the traditional ways that we've done things are sometimes the most elegant way.
I did a an interview back in twenty sixteen, and I said, beware of hipster tools.
You know, as as networkers know, if you don't know how something how you shouldn't use something, you don't really understand enough to use it.
Everything has limitations. So Oh, yeah. You know, find the right set of things. And the good news is, again, there's a lot of training and and resource and data out there, and a lot of people actually even in the industry, you know, that are happy to help.
There's NANOG and AutoCon. There isn't really interop anymore, but there's a lot of resources, you know, out there about how to do it and templates and people that provide services to help not run it for you forever, but augment your team and do training and and help you because it really is a lot more stable, a lot less hard, and a lot more cost effective, especially in terms of of protecting the revenue of the business to put a diverse set of, networking elements, and services in place and have an understanding of how to run those, in the hybrid sense. Right? Your own infrastructure, WAN, SD WAN, cloud data center, use of the Internet, CDNs.
You can get a a great bunch of vendors, that give you those those things that are not trying to control your life and destiny. And, you know, even including the cost of people can be incredibly cost effective.
And then in turn, actually control your destiny.
Right.
In the sense that you have that diversity of vendors Yeah. Without the overhead and and having a a new puppy to take care of, like you said, Jezzibell. Right?
So I would like to talk about, some literally, as you're speaking, Avi, on my left is my Slack window, and I'm seeing alerts about new, security breaches. And so inevitably, my mind goes to, this is nuts. Every day, I'm reading about one, two, three new security breaches, and I'm not a security person either. So, Joe, I'm thinking about you here.
How does, an operations person, an engineer of any stripe, how do they how do they take control of this overwhelming, like, flood of constant security alerts, breaches? And I don't even mean necessarily alert fatigue. I just mean with the things that we have to think about, as we move data across the Internet. Right?
That's, you know, ninety percent of the applications that I use are somewhere else and not on my my laptop here. And, of course, we utilize things like public cloud and now, publicly available large language models like GPT and so forth. So how do we do do we just buy more firewalls?
Yeah. That that's, again, that's that's that's a a fantastic question in the sense of, the your users, the the enterprises are ahead of the operators from that perspective, so you're playing catch up a little bit. Unfortunately, firewalls imply back hauling, because you can only be in so many places, and now you have users because of COVID and other things. They're remote more than ever.
The, you know, the the offices besides some of the some of the big boys, the offices are are shutting down or they're they're they're they're, hoteling. And so, anyway, so I guess to answer your question is it comes down to which is a common theme in control, which is that visibility. And so, shameless plug again for Netskope, a cloud based security service that that can give you that visibility. And and one thing that's interesting is the network person doesn't always consider, security solutions as an additive to their life.
Right? So, a security person would deploy as many applications as possible as long as nobody noticed. And the network person's, I think the joke was the network, operator speeds up packets and the security operator slows them down. And so, and so with, and the and the the thing that I like to tell them is the security people is that if you have a performance problem, you have a security problem.
So if you're on a managed device and you're running a client or an agent or you're back hauling and it's slow, you turn it off and you go direct to Internet or you go to your own, you know, bring your own device. And so as a network operator, you have to make sure that you have a performant infrastructure. As a security operator, you have to make sure it integrates with the network. And so, you know, that's a a long roundabout way to answer your question, but it comes down to having visibility, understanding what's happening on your network, goes back to that layer of control of infrastructure.
If you can give a clean, highly performant path to their destination, that gives you options from monitoring, from filtering, from running a firewall. But if you, if you're running over an MPLS network backhauling to a central data center using a third party firewall, and you have performance complaints, how do you unravel that? Right? And so simplify it, strip it down, get it to the to core physical capacity, and then and then build up on top of that.
Right? If you ask Avi to take over an enterprise, if, you know, if he ran Kmart's network, right, the first thing he would do is is build and build network or get IP connectivity and then work from there. The difference here is it's the network is usually one of the last things the operators and the applications think about, and that's that's, I think underemphasized when it comes to the enterprises, at least that I've run across. So I don't know.
Avi, what do you think? Yeah.
I mean, from a security perspective, having an inventory, having the telemetry this is a telemetry now podcast from the different layers of your network, the things you control and the ones you don't. So your data center, the clouds that you use, the Internet, although that may just be performance telemetry because, you know, you can't get that kind of telemetry from others, Sometimes even down to the host, which can give you great performance data, making it more useful by enriching it with your systems of record, sources of truth, what application, customer group, Those are also really important for, you know, automation if that's what you you know, as part of as part of increasing the ability to operate and increasing security, and then understanding and sharing, as we talked about in other podcast, between the operations and security groups so that when, if the security group has implemented some policy that's affecting traffic and, you know, you don't waste time trying to figure out whether it's a network problem or not, and that the security group has that kind of visibility, Phil, as you were talking about in terms of the security streams coming across.
Mhmm. You know, it may be a public website which has no interesting content and, you know, people, pulling some stuff from it from, from, you know, countries and locations where there's actually no users of interest, and you may say that's lower priority or it may be something else. But if you're stuck in the sort of last decade of scrying by IP address and VX LANs and MAC addresses and ASs and don't know what the traffic means, then it's then it's important. And, again, if you if, you know, the wireless is a black box because it's all some controller you don't have access to or the SASE, then that becomes a problem too.
So it's a lot of is designing with observability in mind, getting the groups to use the common tool set. And then, yes, bottom line is there are a lot of security alerts, and some of them are really high priority, and it just needs to be, you need to be able to determine whether the most actionable by bringing all the data together and enriching it, and then, you know, act on the ones, be able to act on the ones with confidence that you're not gonna break things, or that if you are, you know what to do about it.
Yeah. Yeah. Two things that stand out from what each of you said is this idea that many or most, security incidents, they manifest themselves as performance degradation, maybe even a hard a hard down. And I know that's not always the case when you're, like, piggybacking something in DNS and do some kind of data exfiltration.
You don't want performance degradation because then you get found out. But a lot of the time, that is the case. And so that visibility into, the application delivery system, all the components involved with the network, you know, in the the broad term, is incredibly important to getting a hold of and getting control of your mind wrapped around your security environment, your your how how safe is this data in transit? And then, of course, we can talk about, you know, data at rest.
And so also thinking now about what you said, Avi, which, you know, in my world is well, my world up until about five years ago when I was working in, in in the trenches is the fact that, like, I have so many disparate systems. A lot of them, I own. Many, I don't. And, you know, in order to understand, like, where there's a problem and, you know, where there's a vulnerability and where are the attack vectors and, you know, with all that.
It's just so difficult to put all that stuff together, let alone put it together manually. Right?
And then, you know, finding some sort of context to layer over that. You know? Like, even if it's as simple as, like, here's an application tag so that way I can know, like, you know, the this flow and this metric and then this, like, you know, this this firewall rule, they all kinda relate to each other. Right?
Something as simple as that. That that was struggle for me is, like, just more and more systems that all fall under this umbrella of the, quote, unquote, network and, therefore, like, under my purview to figure out and under my purview to secure. That that was a struggle. And so, yeah, I see, network observability, the the a lot of a lot of that involving the unification of a lot of the data so we can have a holistic understanding of of this system.
So Yeah.
I think that there's two trends that neither of which is the complete picture.
One is the the the old Star Trek NOC. Right? Every every carrier, John, I know Alkami was guilty of this too. Did you have, like, the Star Trek NOC that you would bring people onto the bridge?
There was lots of screens that no one actually that wasn't actually what anyone used, but it had pretty pictures and television programs and It was good for sales to walk customers prospective customers.
Exactly. For sure. As all those systems, as as as you said, Phil, they had different views of what is an application, what is a customer, what is a whatever, you know, looking at elements, but not composed services. The other side, everyone's wants the spog, a single pane of glass view, but there's only so much workflow ish stuff you can do in those because, you know, typically, your Grafana plugin is not gonna have the full functionality of your application or network or or all those things.
So, really, the key is that all those systems that you're gonna rely on, both you need a high level view that people use with insights and and and the proper amount of alerting. And the systems that you're going to rely on, whether they're homegrown, you know, or or commercial, have a common way of talking about what the services, applications, networks are for your infrastructure. You know, that's gonna be, again, the best way to accelerate the humans and reduce downtime, you know, improve capacity planning, reduce cost, you know, all those things. And so it takes a little bit.
It's not that hard, but, you know, still the biggest effort I would recommend for people in the New Year is, you know, if you're gonna make an IT, or even security and or or both resolution, it would be to, be able to answer the question, you know, where is your source of truth?
Do do you do you know where your packets are?
Yeah. And and metadata. Right? You know, I ask people, you know, I I go in and I I say, you know, so you know where you know, you could describe to me what all your applications and assets are and where they are. Right? And, you know, they're like, right. That's a good one.
But, you know, that's what we all need to work on because these systems can't be maximally effective in team.
I always joke that, every single network enterprise person I've met with has inherited someone else's mess, and that person was a moron. And so everybody's stuck with some legacy infrastructure, some Windows NT device, or some architecture decision that that they're trying to unravel, and it's just ties into the do you know where it is? And do you know where the data is? And, you know, that's why we've been pretty consistent about control visibility and, and the market and the technology has evolved. And I don't think it's gonna cost you any more money if you do it the right way. So there's definitely a path out of there, but you you you gotta own it. You you you gotta own that own every aspect of it if you're gonna wanna get it right.
And you just chip away at it. Right?
Start Yeah.
Good. Like, I I came into a technology company, and, you know, they had a list of where all their servers were that was in a database, and they had a file that lived on the servers that said had the list of the servers. And someone came to me and said, I want your people to, you know, like, every week reconcile these things. I was like, why the hell do we have like, generate one from the other.
That's what you do. Make one lead. Oh, but neither of them are completely correct. I'm like, well, guess what?
The thing on the servers that they use, I'm gonna go with that. That's my executive decision. Generate the database from that. Oh, that's stupid.
I was like, okay. Delete the database, generate it from that, then we'll use the database, and then generate the other way. Like, just pick something, solve a problem in a domain, and then progress. And, eventually, you'll know what your stuff is, which is really critical for all this.
So Right. Security, automation, and making the humans more efficient.
Right.
Well, that's, really interesting that it did come back to visibility, because, frankly, I mean, that makes a lot of sense. Joe, do you know where your packets are? I mean, we're talking about, running networks that are systems of systems, and we're not talking about the Internet necessarily, but we're talking about wireless networks and, you know, overlays in our campus and our data center. That boggles my mind that we're running them in campuses now. We're talking about, endpoints and home networks.
So, you know, how how do you reconcile that? It's by understanding the flow of of traffic over all of those things. So so, Jezzibell, Avi, and Joe, thanks so much for joining me today for this episode and just kinda taking a look at how folks can take control over their own operate operations going into twenty twenty five next year. So for now, if you have an idea for an episode or if you'd like to be a guest on Telemetry Now, I'd love to hear from you. You can reach out to us at telemetrynow@kentik.com. So until next time, thanks so much for listening. Bye bye.