Welcome And Guest Introduction

Tim 0:13

Hey everybody, and welcome to another episode of the Cables to Clouds podcast. We have two co-hosts with us this week, uh, Catherine and myself. And we have a new guest who hasn't been on the podcast before. So we're gonna go ahead and let him introduce myself, uh, Tom Kunath from Cisco. Tell tell us a little about yourself, Tom.

Tom 0:34

Tim, thanks. Uh, appreciate the opportunity. Um, yeah, I'm Tom Kunath. Um, I work at Cisco, been here for about 23 years now. Um, my whole career pretty much uh focused on, you know, kind of core um routing, switching security, right? And um with a with an emphasis on, I would say SDN solutions, right? Uh kind of starting with um, you know, this technology called optimize edge routing, which was, you know, application aware, um, kind of morphed into um performance routing, which then became um, you know, I WAN, the IWAN solution. So I did a lot of work with that, you know, wrote a lot of the CBDs, did customer design work. I worked in um professional services for about 15 of my 23 years before I came to technical marketing. And then when Cisco bought Biptela and we had a proper SD-WAN solution, or our first one, um, I, you know, was um one of the first engineers to work on that and camps and then start to do CBD work and um and that kind of led into um my work with satellite. So um, you know, a couple of years ago, um, there was an interest in uh uh kind of forming a loose partnership between um Starlink and Cisco to you know kind of figure out a better together story. Well, like when Starlink goes into enterprise sites, Cisco is almost always the incumbent, right? And so the first question is what are the best practices? You know, how do I how do I leverage this new transport um for either connecting the unconnected or for you know resiliency? And that's where you know we we we did a joint um equipment exchange and I, you know, set up a lab in in RTP and done a lot of testing,

From Military Satellites To Enterprise

Tom 2:10

and I'm now just about to um release a Cisco validated design for Cisco West D WAN with Starlink. So that that's kind of my background and what brought me here today.

Tim 2:21

Awesome. And uh yeah, so wow, 23 years, Tom. That's crazy. I didn't realize.

Tom 2:26

Yeah. Yeah, it's kind of interesting because when I started my career in in uh networking, it was actually even before that, I was I was in the military. I was a um signal corps officer in the army and uh Desert Storm era. So uh one of my first jobs was you know working with a geosynchronous satellite uh in Saudi Arabia. Like, how do we bring cable TV down to uh you know one of the princes?

Tim 2:47

And uh some really important stuff.

Tom 2:49

Right, right. And then, you know, um beyond that, obviously there was, you know, setting up you know secure communication centers, but this still like it was just transport to me at the time. And my, you know, I was my focus was more the you know securing the crypto part of it, but now it's kind of come full circle. Now we have these, you know, low earth orbit satellites that are you know commercially viable for enterprise traffic. You know, back in the day it was teletype, you know, it wasn't real time, uh, it was uh observation and and all that. But now it's you know interesting that kind of a you know bookending my career here um that you know it's it's relevant again.

Katherine 3:24

That's actually pretty fascinating. It must be uh quite the uh the you know, quite uh amazing to watch the evolution of like satellite going from like, you know, from just like very basic applications of military to, you know, commercializing. Like what's over your course of your career, what what do you think is like the greatest advancement or change you've seen that you've kind of enjoyed seeing or working with?

Tom 3:48

Yeah, I guess it really, you know, kind of started, and I lost track of it to be honest. I mean, you know, it was um, you know, military communications. I mean, it was back when we had like 2400 bod, you know, modems and you know you had uh, you know, I'm kind of dating myself.

Katherine 4:01

But you know, uh when I moved to Back when the internet used to scream when you connected to it.

Tom 4:05

Yeah, that's right. When you had these tones that you had to sit through and sacrifice. Right, right. But it, you know, I just had thought it was a thing of the past to be honest and moved on. Uh, you know, really it was, you know, most of my career was like convincing people that the internet was uh was viable enough to carry their enterprise traffic, right? And and then you know, convincing them to peel off, you know, from you know strictly MPLS to go hybrid and trust it. You know, you can run voice over it. And and uh, you know, that then we found that that was the truth, right? That that um you know the internet um service providers are using the same equipment, the same, you know, circuits as as you know, even private MPLS and and and Metro. Um, and as capacity in, you know, the amount of capacity you got was incredible. Um sure you didn't have you didn't honor QOS, but do you need QS? So a lot of my you know middle years were focused on IWAN, you know, giving them a uh an intelligent routing path. Okay, you know, trust but verify, you know, uh sure your banking, you know, traffic can go on MPLS, but hey, why don't we peel off some of the recreational traffic to internet and see how it goes? And oh now that's working, let's peel off some of your critical apps. And and now we can do the steering and we can do the performance monitoring.

Internet First SD-WAN Mindset Shift

Tom 5:20

And and now I think we've gotten to the point where I've seen um, you know, internet first strategies with with even banks, right? Now internet, internet, right? And um and then lo and behold, now we have this uh this this this new transport model with um I call it multi-orbit. I mean, everybody's focused on low Earth orbit, which is the um, you know, so if you look at the broader spectrum, we have um when it's up multi-orbit, we have geosynchronous satellites, and these are the things that I worked with when I was in the Army, right? And these are like 35,000 kilometers in the sky, 500 milliseconds to 700 milliseconds of latency. Um, the beauty of uh geosynchronous is they sit at the equator and they uh rotate the earth at the same speed. So they're always in the same place. You bring you set a terminal up, you point it in one direction, it can always get a signal. So it's great for things like broadcast TV, um, you know, um content delivery, things that aren't latency sensitive, right? Observation, weather, all that. And then there's middle earth orbit, which don't we don't really hear a lot because Leo's the shining star right now. Um that's where our you know, in between. So it's like 20,000 kilometers. Um we're down to about 120 seconds, 120 milliseconds of latency. This is where the GPS sits, right? And uh satellite radio, things like that. Um uh you know, with with Geo, you know, think of a flashlight on a piece of poster board. If you take that flashlight and you hold it out here, you're gonna get a broad circle, right? But it's gonna be not very focused, right? It's gonna be you know, think of that as geo, but if you have like a globe, it only takes three flashlights to cover all that, right? In in different regions of the of the equator. With with Neo, now you're closer. So you need a few more of those. You need like 12 or 25 flashlights to cover a whole Earth. You can get more focused. Um with Neo, you know, the the use cases you get you can get high bandwidth because there's you know, thus um uh you can cover a lot more area. You don't need quite as many, and you don't need to link them together with um with these intersatellite links like Leo. When we get to Leo, now we're down to like 160 to 2,000 kilometers. And and you need like Starlink has currently 10,000 satellites in the sky to cover the whole earth, and they gotta go in multi-pol you know, polar regions and and you know to cover the whole earth. Um, because 80 I say 80% of the earth is still unconnected with respect to internet, right? And you know, we're thinking oceans and and places like Africa. Um so that you know the the um and again the benefit with Leo is now we're down to like you know, they I think they're Starlink saying 25 to 60 milliseconds. Um I'm on a Starlink right now and I ran a speed test right before this, and I got 19 milliseconds, right? Um and yeah, I bought 100 meg down, nine meg up, right? And this is on their old G1 hardware. Um with with a yeah, so so you know, everybody's focused on Leo because that's where we can run our enterprise applications. Um and I I gotta say, it's it's it's it been improving over the past two and a half years since I've been working with it. If if this was two years ago, by now, I would have dropped out of this conference because the there was so much loss. Um one of the problems with or challenges with low Earth orbit is these these um satellites, they travel in what's called a constellation. So and there's many constellations. So it's like think of like one after another after another after another, and they all have these laser inter-satellite links that interconnect them,

GEO MEO LEO Explained Simply

Tom 8:55

right? And and uh the idea now is is as a user, I have this terminal, right? A user terminal or an antenna. Um it's about seven seconds uh for one satellite to cover the whole horizon, right? So you know every seven to ten seconds, I've got to lock on to the next one, right? And and initially there was a lot of loss during this because there there was some mechanical um um you know dishes where they actually you know would move. And then they really evolved that to be phased array now, so they just shift frequencies so they don't have to move it. So now the tracking's a lot better. But during that time, two years ago, um, you know, every time, you know, every satellite, Starlink satellite was programmed to every, I think it was every 10 seconds, switch to another one. And so now you've got kind of like this global synchronization thing where everybody's trying to connect in a region, in an area and a footprint um to uh a new satellite. And so my theory is it was like, you know, there was no tack, there was no connection, you know, like like regulation. So everybody's trying to uh connect at once, and there was some, there was micro bursts of loss. And what we measured, we you know, we we did some early testing with like thousand eyes, and we we measured, you know, over 2% loss, right? And and so uh I think that was a lot of things that were happening together. It was the older technology, um, uh the it was the um just less uh satellites in the sky at that time. I think Starlink had 6,000. So there's just you know oversubscription in the air. And then their ground station, right? Because remember, the the whole path is, you know, you you go up from my my antenna on one wavelength, uh, on one frequency. Um, if I hit that satellite and it can reach a ground station right away, at that same satellite services me. It goes down to a ground station, it backhauls to a um a pop, and then it goes to internet partners, right? And um, and and I believe that at the time, you know, there was a lot of oversubscription and there wasn't like as much thought as the ground, you know, as the ISP engineering as there was in the air. And and so because I was seeing you know, I was seeing like with SD WAN, BFDs flapping, and it was just it was pretty unstable. And like I said, I would have been booted off this this conference by now because there would have been so much loss, but that has um gotten better, and I think it's because of the number of satellites that have increased. Now we're over 10,000, like 12,000, and then the ground state improvements.

LEO Handoffs Loss And Ground Stations

Tim 11:27

So it's uh so can I ask, just real quick, sorry, sorry, Tom, can I ask? Because this is this is fascinating to me, this idea that like every second every seven to ten seconds or whatever, we're basically swapping satellites as if it's kind of like it's almost like wireless roaming, except you know, it's it's the uh the APs that are roaming and not the not the user. What how does the handoff like how does what does that look like? I mean, all you said that all the satellites have like you know laser communications to each other. Is it like like is it like roaming where you would like the APs would like hand off the the signal to each other or like what does that look like?

Tom 12:01

They can, yeah. So so the whole concept of um you know in the air is these intersatellite links. I call them OISLs, optical intersatellite links, space lasers, right? So the um every think of every satellite having, you know, almost like a backbone link, right, to another, right? And um, and before I even go there, like let me let me talk about like before we had these intersatellite links, we had single satellites that would um service uh an area, and they called it a bent pipe model. So think of a mirror in the sky. I would just shoot up and it would just reflect, it would go what was called ground station hunting. It would look for a ground station and it would send it down. If it couldn't find a ground station, you you're out of luck.

Tim 12:42

So if if you're you just don't have connectivity.

Tom 12:45

So that you know, and and a lot of and the the pros and cons because the the um the payloads in in space to do a bent pipe model are simple. You don't have to, it just stays completely analog. You don't have to demodulate it, you don't have to make any decisions. Uh the the the the tech is really easy. And so there's still actually, I think one web is still a type one. Um, you know, you just need a lot of ground stations, you know, built out, right? Um now there's type two, and now they have uh the ability to relay traffic. So if I go up to my satellite, send my packets, I'm in a place where there's not a ground station, it can relay that, it demodulates it. It can, it can say, okay, this is the satellite um terminal. Um it's supposed to go to this ground station, um, and and it can relay it down the constellation across these optical intersatellite links. So think of it as like a core backbone, right? And now you don't need quite as many ground stations. Um, and uh, you know, think of it just sort of as like a you know a core, you know, in in the sky. But there still is the you know the satellite, you know, hopping, um, but they've gotten the loss down, you know, uh uh quite a bit less, right? I I still feel like like when I build SLAs for applicational routing, I still um kind of peg it at 1% loss. Like I say, if you got an application that can tolerate up to 1% loss, you know, that's the SLA, right? And and uh I think that'll that'll get better. Um, but still, you know, one percent loss, 100 milliseconds latency, you know, and about 30 milliseconds of jitter is what my application were running policies in the C V D specify. And that and that should get better.

Katherine 14:30

That's close to I was gonna say that that's close to acceptable or pretty close to acceptable for like uh VoIP applications for enterprise. I ideally they want it under 100 milliseconds, but it can tolerate like 150 milliseconds or lower is generally considered okay. The loss, uh here give or take, but it is getting better. The above two percent would have been a a disaster, but you know, getting uh you know one percent or under, that's actually pretty good. I i it at this rate of that that they're getting better at, I I could see a lot of enterprises being able to switch over to satellite.

Tom 15:08

Yeah. Yeah, I mean, like again, the proof is like I'm on my Starlight right now and I'm talking and you're hearing me fine. There's

TCP Backoff And SD-WAN Fixes

Tom 15:14

no blips, and I mean, I think the video's coming through. Um now the the loss um is interesting because um I don't know if you, you know, you guys probably know. Um, you know, most of the TCP stacks that we have on our computers are using like older congestion control algorithms, like you know, like the original like back off Sawtooth algorithms and stuff like that. Yeah, Cubic and Reno and and the whole like you know, thinking was that if there's um uh you know, if there's loss, it's probably because we're oversubscribing some low speed WAN length. So so the algorithms are like, you know, trying to be very respectful, and and they say, well, let's cut my transmission rate in half so that I can slow down, not congesting that, right? And you know, that's what's built into our Macs and our and our PCs today for the most part, right? And so what we saw was initially when when we started, you know, before I even got involved in this, customers started using Starlink for backups and they stood and the first thing they wanted to do was run a speed test, right? And then, oh, it's not too bad. But then they would run iperf tests, which is like sustained through actual TCP, and it would drop to about 25%, you know. So they're only getting 50 meg out of what Starlink said they should be getting 250 down. It's like, well, why is that? It must be, you know, Cisco's dropping something in IPsec. And so I, you know, I I got really focused on this and and looked into it and um you know, went through the box with escalation engineers, and there's no packet drops, there's no fragmentation, all the normal things you think about. Um, and as it turned out, it was the you know, it was the the stacks backing off their transmission rates because they would see this loss in the form of missed TCP sequences, you know, and it would just say, hey, we got to slow down, right? Yeah, and so um one of the great use cases and and things I showcase in the in the CBD uh the design is um Cisco's app QoE features and and TCP optimization is one that's very interesting, which is like proxy. So, you know, on the router, um, you know, you're you're you're spoofing, it's almost like local acts, right? You're you're telling the clients, yep, keep sending, keep sending, keep sending, there's there's no loss. On the server side, same thing. And in between, we run um uh what Google came up with this BBR2 protocol, which is more intelligent. Um, at some point it will make its way into the into the stacks. And um, you know, it it takes more into consideration than just loss, right? And it can determine that, okay, that you know, maybe there was latency or you know, like this link is not you know congested and you you can continue to fill the pipes. Right. So yeah, TP op is a big thing and forward error correction is another one. Drop a bit, um, the receiving side can use a parity bit and it can reconstruct it. Um, packet duplication is another one where, all right, I've got this critical site. It's let's say it's an oil rig out in the ocean, it's got to get the sensor data back. Let's put two Starlings on it, orient them in a different area so they connect to a different um, you know, uh satellite in the air, and then let's stream the traffic over both, right? Like in just, you know, it's almost like multicasts. Like let's make a replication, send it, the other end of the ST WAN, we'll reassemble them. Figure it out. It'll take the ones, the first ones that come in and it'll reconstruct them, send them on their way. I mean, these things are innovations that have really um you know helped a lot with uh with satellite communication.

Tim 18:41

Yeah, I didn't I hadn't thought about, I mean, the old the old TCP back-off and sawtooth algorithm that you see with TCP is like so old. I remember when I was first studying networking. I mean, and that book, the book I was reading it was old even at the time, like TCP IP Illustrated, uh volume one, which is like the the the book on TCP. But I mean, even then it was it was talking about how, hey, you know, if you miss a packet, miss a sequence, you know, you go to like half speed, right? And then you have to build up again. That's the old sawtooth algorithm. And with something where you're dealing with one percent loss almost as a as a as a floor, right? Like, yeah, you're gonna be forever with this sawtooth algorithm trying to, and you'll never get back up to full speed. So that's really, really interesting.

Tom 19:26

There's a really good paper um that's written uh by a thousand eyes principal. Um, I think of his name right now, but it you can Google it. And it's like the effect of 1% loss. And they don't even talk about traffic, right? About about satellites, not focused on that. But it was like, you know, he just goes into great detail of how it wreaks havoc on today's, you know, protocols. And then he talks about, you know, the whole thing is making a case for BBR on the stacks, right? But um I worked with him, I can't remember the name now, but um, yeah, he he kind of helped us with some of the hyperf methodology testing. And and then, you know, we showed him how like the Cisco as as as a TCP proxy could really help. And um he cited in one of the um articles that I published on the design zone, it's like optimizing um you know, Cisco optimizations for satellite, right? And we talk about a study, that study he did, his paper, and then we we used um Meraki as well to take their own. They have there's a very big um uh satellite you know base with Meraki. And since they are all cloud managed, that you know, they have like telemetry and they have visibility of like you know, the use cases, and they can they can go in and and check the latency and the performance as well. And they were then and at that time it was like almost three or four percent spikes of of loss, right? So but I I think that's come down quite a bit since then.

Kuiper Competition And Capacity Limits

Tim 20:46

So all right. So we talked a little bit about uh you know low earth orbit being the the media darling at the moment because of course Starlink has just got so much press and everything. I mean, there's not just Starlink out there with low Earth Orbit constellations, of course. There's there's new companies, other companies that are doing this as well, right?

Tom 21:03

Yeah, I mean uh you've probably seen the press. I mean, Amazon Leo is um, you know, they're gonna go head to head, right, with Starlink and then Project Kuiper, yeah.

Tim 21:11

Yeah, yeah.

Tom 21:12

Project Kuiper is now like, you know, it's now Amazon Leo. Um they've got uh I think they've got like four or five hundred satellites in the sky. Um it's interesting because you know they're they're they're targeting, you know, enterprise and and even residential, um, you know, same market. Uh they're type two, so they, you know, they they use ground stations and they um but they've they've leapfrogged above in the in the um engineering of the antennas, right? The antennas they have um much higher speeds. Um um they they can get, you know, um uh gig down and four hundred mega up on their higher end. Uh they use um you know just advanced phased array technology to where now they even have like one that can sit on an aircraft and it can service the whole passengers and crews um at at decent speeds, right? So I think we're gonna see that. Like, you know, Starlink has the ability to do that too, but then they have this capacity issue. If they if they turn up the you know the speeds, now they've got uh, you know, c concerns about the, you know, the sky on the satellites yet. Links are like 10 gig going to 40 gig. I think they might be up to 40 gig now, but like there's this like, you know, bandwidth, you know, capacity, you know, that you're always chasing. And now as more, you know, they've got I think a million subscribers or yeah, they you know, that's it's it's just growth.

Tim 22:37

It's the cloud story all over again.

Tom 22:40

Subscription problems. Right. Or even like I I think of it as when people started to figure out that they could use 3G for data, right? The uh you know the phone companies were like, oh my God, I didn't we didn't expect this. We didn't engineer for this and you can't do that. You get data caps and it's kind of the same thing, right? It's like people are finding innovative you know solutions to these problems and and almost you know the providers are you know aren't anticipating them.

Katherine 23:05

It's funny because with uh capacity problems, I think that's like a tale as old as as the internet basically when they first created, you know, it was like AT ⁇ T and a bunch of other companies started uh building backbones for the internet and and ISPs uh they never imagined a day where there'd be streaming video and audio and stuff like that. So they hit that limit pretty quickly and then they're like oh crap and just constantly upgrading. Seems like the same thing with satellite it does.

Tom 23:33

It does yeah and uh and that's where I you know I'm starting to see now um like serious players like you know the I work a lot with mission critical networking so I'm working with the you know the the Army the services and intelligence agencies and and the um the NATO country Ministry of Defenses right and and a lot of um I'm starting to see a lot of multi-orbit designs right where they'll have like you know let's say a ship um and uh ship to shore communications and and they will take the traffic in from SD WAN and put it over like an SRB six backbone and they can they can say okay we want to give you a low latency slice and that'll go that'll be directed over a Leo Leo constellation. Well you might need a high bandwidth um slice that's maybe not for real time traffic. We can send it over a Mio satellite right and and uh and then geo is for you know your traditional um broadcast stuff and and even even high bandwidth there if it's you know not time sensitive right and so uh yeah that's that's where the multi-hor stuff comes in um pretty interesting and and there's a whole nother side that probably for a topic for another um podcast is is this whole concept of non-terrestrial networking.

Non Terrestrial 5G And Direct To Device

Tom 24:46

And uh it you know that's the broader I guess envelope that all this multi-arbor falls under. And then non-terrestrial networking includes um cellular like 5G in space. So so um you know these these these providers like Starlink now have um started to partner with companies like T-Mobile um and now they can offer and they bought some spectrum and now they can offer 5G like almost like a global long distance roaming service right so if you have a T-Mobile satellite with the same chips sets in it that you would have for 5G, if you're in a location that can't get terrestrial, no cell tower access, and you buy this plan, you can pick up like like satellite service from a Starlink um satellite, right? And uh and again it's today it's almost like the Bentpipe model where my phone has the modem it can it can connect onto that signal and then the satellite just finds a cell tower you know ground station right like loose speaking right that makes sense. And and now you know for like remote areas and and very low bitrate stuff. But the um there's a big um I guess initiative going on right now to standardize this and um it's all part of this 3G PP um release like 19 where now they're gonna um you know basically converge um terrestrial 5G and non-terrestrial 5G under the same standards, right? And so they're they're gonna be able to put cellular ground stations on on the satellite. So think of like I talked about type one and type two where type one is a bent pipe we don't demodulate it stays analog we just amplify it ship shift the frequencies down very simple. That's how that's where voice is today with with respect to um this this this um satellite you know um voice um going forward you know there's going to be need to be new chipsets um in in phones and modems they're going to be able to demodulate the traffic in space have the um have the ground station um technology rate in space and and then offer you know much better um speeds and availability so so um you know the that that's the you know um I guess the the big revolution right now going on on the I call it the boy side or or the you know the the use cases there they call our direct to device or direct to sell. And direct to device, you know think of an IoT um substation somewhere uh like maybe it's an oil field, right? Where you don't even have power. Like you're not gonna set up a Starlink which needs external you know DC or AC power and forget about it, right? But you might like be able to set up one of these you know new new radios which are you know they're calling like the sta the standard is NTN IoT, right? Which is you can have a battery that can last a year and that um signal is strong enough to where it won't drain you. And so now I can have enough bandwidth for you know to to relay the the you know the sensor data back to a station and and kind of forget about it. That's NTN IoT. And then there's another NTN new radio standard for um basically regular enterprise kind of communications. And where I am seeing some interest here is like some of the banks and and you know it's a it's really a resiliency play. Like like a lot of the banks are using they have ATM machines everywhere, right? Not just in banks. They're in malls or they're everywhere. And right of course and there's a concern and and they primarily use internet and and um you know um there there's a concern that you know could be a cyber attack it could be you know uh you know bad player trying to you know sabotage uh you know uh the terrestrial um they want some type of non-terrestrial like last resort solution right so initially we pitched well yeah Starlink's perfect you know you put a Starlink terminal on the you know roof uh you you know you bring the Ethernet down and they're like hold on wait wait like no we we've we've already got the cellular um you know uh antennas on the roof and and you know we're using terrestrial cellular already um but even that's vulnerable because you can jam you know like that terrestrial cellular um signal right you can you know you could it's there's still a dependence on a cell tower that's physically on the ground right so that can be still a we want you know we we we've heard about this like direct to cell and we want to be able to you know have you know satellite as as a last resort. So even if like there's bad actors that are jamming you know the 5G signals um we want to be able to do that right and so and and what that for for us it's great because we don't have to you know go and change infrastructure and run you know Ethernet cables and power up to you know in every location you know to a Starlink dish, right? And we want that. And and so we're still oh uh probably I don't know month to not a year before that standard is there and our routers have the modems that have that new um NTN capability but that I feel like that's kind of the direction um that a lot of uh you know our enterprise customers will will want to go is is is you know the 5G NTN satellite right um so but today Leo you know think of it as fixed broadband you know broadband satellite internet that um is a little quirky right so you still have to you know condition it and use these features that we have uh know that it's asymmetrical that's another challenge you know like I said I I got nine meg up and a hundred down um so uh that nine meg tomorrow might

Weather Obstructions And Mobility Reality

Tom 30:34

be better. I mean it's been raining here you know one of the things that um you know satellite frequencies are are susceptible low Earth orbit is is rainfade right and and um you know so you're gonna get a variable you know performance depending on weather um and as well as like if there's obstructions right if there's you know things going on of course and what I found was you know I tried to do some convergence testing uh being in a test engineer I was like all right let's let's see how fast I can get things to converge from satellite to cellular and foolishly like I I was out at the beach I was in a place where you know I I didn't have much good you know connectivity so I wanted to do that and I thought all right I'm just gonna cover it with cover the dish with a surfboard and that should block it right thinking because you know they they tell you you got to have a complete line of sight you know and and it just got working I was like wait a minute so then I went and I stuck it in my truck covered the the the the to no cover and it was still you know it didn't convert I still had a great signal and then I realized it was the physics of it right that those those um KU waves the frequencies can can penetrate um those types of materials metal and water right those are the things that really um create you know wreak havoc mess with those waves yeah okay yeah so if you're like you you probably have seen this like you if you use satellite serious satellite radio you know you have a signal everywhere but then you go to a drive thru to get coffee and it clips out because you're under a bell awning right um or another one is like I I started to play around with like you know a little bit of mobile mobility right so I was like all right let's just you know we've had some customers that in the utility space in the military space that want to use like Starlink plus satellite right and and so I've been trying to you know again figure out okay what's the you know application we're routing you know how can I monitor both and switch over um and uh you know put it in my truck and and said okay let me let me just like stream my you know music uh on the satellite let me try to take some calls let me do some you know and then if I want to do some like real work I'll pull over and I'll I'll do some work and test right and and when you pull over in the south you know what do you do? You you typically park in the shade, right? And and that shade is under a tree and that tree's filled with water and all of a sudden you know the the the signal fades right so there's there's quirkiness that you need to be aware of. And that's why I don't feel like it's a complete replacement for for cellular in all use cases I see it as a complement right so like satellite plus cellular mobility and transportation is is probably always going to be a thing.

Katherine 33:14

It's actually pretty also pretty good as like a fallback as well if you're you know cell towers or like power outages happen and you need to natural disaster or something. Yes if you need a fallback that doesn't rely on the power like the local power grid and you have generators for your own like buildings power whatever satellite.

Tom 33:32

Probably still the primary use case right is is this you know um resiliency play and connecting the unconnected where you just can't get anything else right um there still leads with backup resiliency solutions because it's just been you know a thing for kind of ubiquitous yeah yeah and and but I I feel like with the advances of of bandwidth and as like more and more AI applications are demanding much more bandwidth the the you know the physics of satellite communications are going to be able to keep up more than you you'll you'll see with just pure cellular, right? So I I feel like that's gonna change over the next five years. And then and then something we haven't even gone to is this whole concept of orbital data centers.

Orbital Edge Computing And Satellite Lifecycle

Tom 34:14

You know there's a vision um well yeah not just them there's Blue Origin there's there's quite a few right and that the whole argument is that um you know we we need you know data centers of space right it solves a lot of the problems we have with terrestrial data centers right and and uh and so Starlink is investing heavily in this concept every every satellite they send up now has like you know half a terabit of of storage and in computing right and and uh and so yeah there's there's you know this thinking that especially for AI traffic where you know you might just have inference you know decision like it never needs to reach reach the earth right you can get you know um send it up to one data center maybe it does the you know maybe you've got like a um drone that's doing surveillance for an area and you send it up it sorts through all the data oh yeah this is a foe like this is an enemy right and this is the type of tank he has and this is what whatnot um and then you know send that over to another you know the metadata of that over to another um satellite for storage because like some you know some of these providers are gonna lean on the compute and some will have storage right so there's you know but that but only the the filtered results need to come back down to earth. So you know the it it it's you know the claim is it's gonna solve the the power issue because you've got unlimited power with with um solar. With solar the challenge is the heat dissipation. You think like hey we're we're in space it's cold but the space is space is a vacuum there's nowhere to there's nowhere to pull the heat away from the chip. So so the the concern is you've got you know today GPU chips that you know you've got to cool um to you know and so what that means is you've got to have these huge um radiators on these satellites that have you know basically chillers and they're pulling they can actually pull the heat away from the chip and then send it into space. And that's that's the big challenge that and radiation because the the radiation will cause bit flips and and I think today like Sam Altman says like forget our data centers aren't even viable like let's not even let's not even have that conversation. Elon's say indifferent um the industry is investing heavily in in in it so I I feel like it's a problem that's worth solving and eventually it will get solved. But yeah the the radiation is a problem um uh and and I think they're saying like you know an orbital data center you know with you know I guess the ROI is about five years or not the ROI but the lifespan right whereas a terrestrial is about 15 years you can get 15 years out of the same hardware five years. So so right now it's you know is it viable I don't know but I I feel like that's the you know interesting part right because if if it does become viable um then now we have an opportunity to do everything we do with the terrestrial everything you've done Tim with like multi-orbital you know our multi-cloud networking right you know think about you know we're gonna have all these siloed orbital data centers and how do we interconnect those and I feel like that's where you know companies like Cisco at least the technology um can play you know kind of like the middle mile optimization right so right it's really an interesting kind of and and they've we've listened to like Tutel like on um I think he did a CNBC um interview not too long ago where he talks about future um orbital data centers and and the intersection with AI and it's it's a pretty interesting listen. And um yeah they're there it's more than a science experiment right now.

Tim 37:53

You know it's it's it's me totally invested in the it honestly reminds I mean just I want to segue into the life cycle thing because uh that you put brought up a good point. But honestly I don't think of it as orbital data centers. To me it feels more like edge computing because we're we're taking all these we're you know 10,000 satellites with all the you know distributed compute distributed uh data and we're filtering the results before we send it back to something some some central repository. So to me it feels a lot that and that's not to argue I'm just saying like just based on the stuff I've seen about it I know they're saying orbital data centers because that's what people know and it's but it feels more like edge compute to me.

Tom 38:31

Absolutely it's a bunch of small compute um you know distributed and storage.

Tim 38:36

And yeah I I would think of it more you know you'll hear orbital data centers but then you'll hear uh orbital edge computing which I think is is a is a more more accurate yeah no no t yeah I'll talk about the life cycle because this is this is something I want to get to I mean 10,000 uh 10,000 plus whatever uh satellites in orbit like I mean doesn't that mean like how do you that's like you an entire life cycle management like nightmare, right?

Tom 39:04

Like what does that look like? Out of these 10,000 like no well the first of all, what changed the game completely like why we're even talking about this now is is the is the is the rocket technology, right? The the you know the the ability to you know launch rockets and have them return to Earth, right? And so that that cut the cost from like $54,000 per kilogram for a for an orbit for a um orbital payload down to like $2,600, right? And so you know at any given time out of these 10,000 they're not all operational. A lot of times they'll take them they'll take some out of service they'll do maintenance on them you know everything has to be engineered for remote maintenance. I mean the these satellites do have these docking stations that can can be serviced in air um and I don't know too much about this so I'm not gonna talk too much about it but yeah um that there will be you know the life cycle at some point that they just will let them burn up and and they'll replace them with new ones right so you know when I think of um designing um you know these it's a lot of high availability right if I were to design a Cisco space routing solution I would have Google everything right and I would have them serviceable and and you know um everything containerized right you know so you so you're you're minimizing the footprint and the power draw.

Security Models And Private Starlink Options

Katherine 40:22

But yeah there's that's a whole topic for somebody that knows the space aspect a lot more than me out of curiosity uh switching topics a little bit um what do you see regarding like security or uh or uh security concerns for customers with satellite because I mean at the beginning of like wireless being a thing like just general like wireless and enterprise people were like first thing they thought about was security concerns. And we've seen like I've seen like normal non-tech people get concerned about the space uh Starlink and stuff like that. Like during the elections there was a lot of conspiracy theories and and things about Starlink and how secure it could really be but my under my general understanding is that like Starlink and a lot of these satellite ISPs like they use heavy duty encryption that are that is not like TLS heavy duty. It's it's like you know quantum safe like pretty hard to crack like constantly cycling huge keys. What it was so I guess the question because I'm not this isn't I'm not super in depth on on satellite security but like what concerns do you see and how comfortable do you feel about the security of these satellites?

Tom 41:34

Yeah I mean from from my experience like enterprise treat it just like the dirty internet right so any you know the the cybersecurity policies they have in place to use internet transports they they apply with satellite right and um uh you know it so trust you know yes that they're doing their own encryption um from the last mile between your satellite dish and their constellation but um everybody almost does IPsec tunneling over right because it's you know again that it it goes up to the satellite and cross intersatellite link then it comes to a ground station and then it gets handed off to the internet providers right so um you know most enterprises are are fine with that they use internet now they they tunnel you know SD-WAN right and and that's the primary um that's why I just published a uh um a Cisco validated design with SD-WAN because that's the predominant use case that's accepted by almost everybody. And I say almost everybody, the mission critical military and and um CNI are the critical national infrastructure folks, the energy transportation grid operators, they say no, like we're not going to use it. We have like double secret security where you've got to have two firewalls if you can use the internet and we don't want that. We want private. And so there are services. Starlink has a a um a private a PI service right because again the concern is ground station to internet right and so for a price you can get a Starlink PI where they program your dish to not just use any ground station but to point to a specific ground station that has a private circuit um that connects to a Starlink pop and it hands off an L2 VPN right so now there's no layer three it's a layer two service. So think of a Metro Ethernet connection in this where you have an end L2.

Katherine 43:31

That's actually pretty cool. I actually didn't realize that they can do that. I I just know that there was like a lot of uh people like not technical people thinking that like you could just hack the satellite uh and I knew this was very unlikely um but hack the satellite be man in the middle right there and I was just like yeah it's not how that works but um I was curious what your your your impression was on like that security and that's cool like the P and I part I had no idea that they actually offered that Service.

Tom 44:00

Yeah, I I wrote a um knowledge article that's out there. If you just search on um Cisco, Starlink, L2VPN, um, I talk about a use case. And this is based on a real customer that was in the energy industry that came to found me and said, okay, we want to do Starlink at these substations, and they were like they were like a transmission energy. So like, you know, there's you know, there's a whole like hierarchy of energy distribution, sort of like networking, right? Where you have access to the houses and then you have like the the middle mile, which is like the the the big, you know, uh transmissions. Though their cybersecurity policies are just as strong as anybody, because if you hack that, you've taken down the grid, right? So they are pushing us. And when I talked about Starlink and the internet, they said, stop, like we don't we can't do that. Like, you know, even with IPsec. Um uh but you know, do you is there a private solution? And then you know, we found that yeah, there is. They have this P and I service. But you know, what that requires is that you know, all these P and I's meet at a colo somewhere, like an Equinix or a, you know, um and like a Starlink rack that um terminates the L2 VPN to to the to the thing. So you've got a completely private thing there, and then you peer with it. And I think the entry point is a 10 gig peering, so it's not gonna be cheap, right? And then you know, from there you can hand off to your own MPLS backbone.

Tim 45:20

Yeah, some other private circuit. Yeah, exactly.

Tom 45:22

Yeah, and it's completely end-to-end. And that's that's today's private networking, but there's something that's coming that I think is even way cooler,

Gatewayless Private Networks In Space

Tom 45:29

right? And it's it's um this concept of gatewayess satellite, right? And so again, everything I've talked about has been kind of last mile satellite, right? We're still connecting to the terrestrial either internet or a private circuit, but we got a dependence on some ground station somewhere. And if you look at the details of like L2 VPN, to me, there's still areas that are suspect, like like the this this private link between, you know, the ground station and this pop. Like, how private is that? Is that encrypted? Is like it's out of my control. So, like this, there's new companies like Ravada Space Networks that's coming out. And what their um whole model is, is no ground stations connectionless. So think about a Metro Ethernet in the sky where you have a satellite constellation and multiple of them with all these high-speed intersatellite links. The only way you communicate is having a satellite dish. So, so like you have to have a Ravada satellite dish at every site. And now you have a private network, right? So they're like, an NPLS provider. And actually, they're they initially they're gonna run NPLS VPN, and now they think they're moving to SRB six, like because it's the you know, kind of mixed up.

Tim 46:44

That makes sense for that for that use case.

Tom 46:47

So now like this, the the whole you know, sovereign networking and mission critical networking um crew loves this idea because now I can have um you know a completely air gap, literally air gapped, and figured to the air gap network, right? That I am under control. It'll never touch a ground station. I'll get my own VLANs handed off to me, uh like a trunking. It it complies with the MEF 3.0 framework. So it's you know, it's it's um, you know, standard stuff, right? Um on the ground. And I have like a peace of mind knowing that, you know, I've got my own L2 VPN essentially that that will never touch the ground, right? And so this is where I think we're gonna see they call it like orbital end-to-end networking or, you know, um they Rivada calls it the outer net, right? And they they came up with the term the outer net, right? And it's uh um it's pretty cool. And I feel like that's the um we'll see more of that. And Rivada's supposed to go commercial by the end of next year. And I've I've written some white papers and potential use cases for Cisco that it's also out there that you can you can look at. And I'm really tracking that um for Mission Critical.

Tim 47:53

Uh-huh. That makes sense. No, this is great, man. Like I would I would love to dig more into it, but uh, we're coming up on time here. Uh I'd definitely have to have you back to talk a little bit more about the broader idea of non-traditional networking. So I think that's a really, really good thread to to pull on for the for the next time we get shown here. But thanks for uh coming on. Where where can people find you?

Pricing Takeaways And Final Wrap

Tim 48:14

Like on the internet?

Tom 48:15

Like um, I you know, I guess just um through my email. Uh no, I I I've you know if you Google my name, Tom Cuniff, you'll probably find a lot of my Cisco Live um breakout sessions, and they're about, you know, primarily you're gonna see SD-WAN um breakouts, but uh more and more they'll be cellular or they'll be satellite. But yeah, just uh Tom Cuniff, Google me and and then reach out.

Tim 48:41

All right, we'll get we'll get all that in the show notes. Um yeah, thanks for joining us. And uh any last thoughts, Catherine? You should wrap it up.

Katherine 48:50

Um, it was a really interesting conversation. I actually learned a lot about satellite networking myself. I mean, it's not something that professionally I've had to jump into yet. I knew you know some general stuff, but that PI service, the non-tra uh the actual like closed circuit, that's really awesome. And it's it's crazy to see how far we've come since you know, the Gulf War where you started to now, uh and just like so much innovation. I mean, and then I feel like the last five, ten years have been like a snowball effect of just how how how how much we've made leaps and bounds.

Tom 49:27

Yeah, I think it's gonna change like the whole Moore's Law. Like I think it's like every six months you see like reaching.

Tim 49:32

Yeah, Moore's Law.

Tom 49:34

And uh, you know, like just again from two years ago to now, the quality, like I I you know, I wouldn't have suggested to anybody to use Starlink a c a couple of years ago. But now here, you know, here we are, I'm on pure Starlink, I use it every day, and I forget that I'm even using it. So I I think that I'm awesome.

Katherine 49:50

I actually looked it up while we were on this call. It's funny how like how it's actually become incredibly affordable from a consumer perspective. I uh so to full disclosure, I've had like the same cable internet provider since like I moved in here in 2021. And uh I've it it it started out at like $69.99 on a promotion and now it's like at $159.99. And it's it's like maybe 200 megs download. Um I go I went to yeah, I went to um uh Starlink's site and for their max plan, which is over 200 and all sorts of stuff, still cheaper than this, it's 130 a month.

Tom 50:28

And is the equipment cost? Have you checked that out? Because initially there was a high.

Katherine 50:32

Oh, it says zero uh equipment cost on there.

Tom 50:34

Right, yeah, they've got to pay promo, yeah. Nice. Yeah, yeah. That that and that's changing too. Like I have the Gen 1 dish. At the time it would have been $2,500, you know, just to get the dish. And and now that's come down, I think, to probably part of the service, right?

Tim 50:50

Probably subsidizing it to get it out there, yeah.

Katherine 50:52

And free professional uh installation for the max plan, which is awesome. And so that's only $130 for you know, up you know, potentially faster speeds. I'm sure the upload is gonna be not as great as cable. But but uh it I I get an outage every like other week here for my cable and have to switch to my mobile. So it's probably gonna be more reliable.

Tom 51:14

Yeah, I live I live in a place called Carolina Beach, North Carolina, which is an island, and and we have spectrum and ATT, and occasionally, you know, it it'll go out. And and the Starlink has saved my maybe must save my marriage because we've had weekends where you know you rely on internet for TV and everything. And um, you know, I just jump to the Starlink and and we're good to go. And I would say if you you know if you want a you know low-cost entry into it, you know, look at the Starlink Mini, which is a portable one. It's what a lot of the overlanders use and the people that are, you know, go to places and and uh, you know, the I think you get them for like, you know, Best Buy for the for the unit, and then it's like a $50 plan, and you get you know, you get decent speeds. I mean, I've used them in the mountains and places where I had no other internet, and they're they're pretty cool. I know um we've got a guy at Cisco, um, one of our product managers, that's he he does like Baja racing, and and they've you know they've got a solution where they've got Starlink Mini on top of the car to stream live and then to send telemetry over. And um that you know again, that that's a portable thing. You don't need a dish. You just self-provision it and it's pretty cheap. You can turn it on and off when you want. So that's yeah, and then at least came out with a mini two, which is like you know gonna be even better.

Tim 52:31

So don't we're we're not being uh just to be clear, uh, this is not a sponsored episode. Uh although Elon, if you if you're watching, go ahead and uh give us a little bit. If you don't mind, we'll we'll be happy to talk all about Starlink. All right, everybody. Let's go ahead and wrap up. Uh thanks for again, thanks for coming, Tom. And uh this has been a great episode. And everyone, on the next episode of Cables to Clouds. Take care.