Watch on YouTube – The Internet Report – Ep. 18: Aug 3 – Aug 9, 2020

This is the Internet Report, where we uncover what’s working and what’s breaking on the Internet—and why. On this week’s episode, Mike sat down with our guest, Ray Hunter, a senior network consultant at Globis in the Netherlands, to talk about traditional SatComms, how it differs from newer satellite technology, and what effect the mass deployment of Low Earth Orbital (LEO) satellites will have on networks and service delivery. We also discuss a recent move by the US to ban financial transactions between TikTok’s parent company, ByteDance, and US citizens, effectively removing financial incentives to serve users in the United States. While not an outright ban, it does raise questions about how an outright ban could even be enforced, and what that means for the broader conversation around Internet sovereignty.

Find us on:

Finally, don’t forget to leave a comment here or on Twitter, tagging @ThousandEyes and using the hashtag #TheInternetReport.

Catch up on past episodes of The Internet Report here.

Listen on Transistor – The Internet Report – Ep. 18: Aug 3 – Aug 9, 2020
ThousandEyes T-shirt Offer

Follow Along with the Transcript

Angelique Medina:
This is The Internet Report, where we uncover what’s working and what’s breaking on the Internet and why. This week, I’m joined by my colleague, Mike Hicks. He’s coming to us from Perth, Australia. So welcome Mike.

Mike Hicks:
Thanks very much. It’s great to be here.

Angelique Medina:
So on this episode, we’re going to talk a lot about SatComms, and we also have a great guest on the show that Mike interviewed, and that’s going to be really interesting to listen to, so make sure you stay tuned for that. But, first up, we wanted to address some of the major headlines, and one of them had to do with TikTok, and the U.S. Government’s announcement that they were effectively going to remove financial incentives for TikTok to continue to operate in the U.S. because it’s going to, in 45 days, ban financial transactions between TikTok’s parent company, ByteDance and U.S. citizens.

Angelique Medina:
So when that does take effect, that’s really going to remove any reason for them to continue to operate. While it’s not outright banning them, because they’re not going to be able to, for example, have any transactions where they’re paid for advertising, for example, from U.S. companies, then there really wouldn’t be any reason for them to continue to stay here.

Mike Hicks:
Well, not in it for the good of the people essentially.

Angelique Medina:
That’s right, that’s right. But let’s say for the sake of argument that the application itself or the service itself was banned. In terms of how the U.S. government would do that. We can look at other instances in which, for example, a website or an application maybe is infringing on the trademarking or intellectual property, or it’s been found to have illegal content on it, or there was fraud involved. There are a number of things that the authorities can do in the US in order to shut down a site.

Angelique Medina:
One of the things they can do is they can go to the DNS registrar or the domain registrar for the domain, like in this case TikTok, and basically say, “Hey, you need to revoke that domain.” Or they could go to their U.S. vendors, like their CDN provider or whoever’s hosting them or their DNS hosting provider, like Akamai, for example, serves their content and go about it that way. There really wouldn’t be an incentive to do it from a network standpoint.

Mike Hicks:
Yeah, everything is technically feasible, but it becomes how much effort you put in to achieve that.

Angelique Medina:
Right. Exactly, and even say if they really wanted to go rogue, and they wanted to serve the application outside of the U.S., and U.S. users wanted to go and access the application, if they have to, especially because we were talking about video, if they’re not connecting, if U.S. users are not connecting to the application in the U.S., so they’re going, let’s say to Europe, or to another part of the world to connect to the application, then their performance isn’t going to be very good. It’s not going to be a very enjoyable experience, and for that reason again, you’re probably not going to need to take any further action in order to prevent US users from using the service.

Mike Hicks:
Yeah, exactly. It’s got to be that user experience. In the early days, everybody started setting up these little server’s oil rigs and the independent domains, but you lose that whole CDN aspect. So everything’s not close, so I’m making these huge journeys, everything looks like I’m in Perth trying to get out to get some decent network connectivity. So yeah, it becomes not worth it, exactly right.

Angelique Medina:
Exactly, and preventing users from reaching a particular service, like for example, what they do in China. There are basically three major service providers in China and heavily regulated versus thousands in the United States, and they all have their own ways of operating their networks. So while it’s technically theoretical to do things like IP address blacklisting or intercepting DNS queries, it’s not foolproof, and it’s not really feasible or practical at the scale that would be required.

Mike Hicks:
Yeah, exactly. And ultimately what comes back to the main thing, take away the revenue source or that becomes unnecessary.

Angelique Medina:
Exactly. All right. Okay, so moving past the technical possibilities or feasibility with a ban like this. We were talking a few weeks ago about the SpaceX’s Starlink supposedly low-latency, low-orbit satellite system that is currently wanting to bid for the FCC’s high-speed rural connectivity project that they have going on in the U.S., and you had talked to someone who had previously, he’s been in the networking industry for over 20 years, and you worked with him in SatComms at the European Space Agency.

Mike Hicks:
Yeah, that’s right. Ray Hunter was talking to that. So we were having this conversation, and we will cover all of this there, but one of the things that came out was that we were talking specifically about the styling staff and the low-latency aspect. And it’s totally feasible to do this. Everybody can have a low-latency link, we can do it between point A and point B very quick. We can have small packets, we can get this fast lead there. But one of the points where Ray brought up was that’s irrelevant, everything you talk about from low-latency has got to come from the application, ultimately, for the end-user experience. So it starts getting into all those aspects. So it’s interesting, and there’s other things that start to impact us as well when we start to get into this wireless SatComms, whether it be geostationary, whether it be microwave links, which is a form of that type of connectivity. And we’ve started to have other out outside influences. So, whereas, we’re dealing with an ISP, a backhoe can dig up a fiber link and cut that connection across the net, it is extreme. It does happen, but it’s extreme there, whereas we’re impacted by things like the weather when we start talking about the satellite connections and the environment, the sea and whatever takes in place there.

Angelique Medina:
Exactly, and I know you guys are going to get into a little bit the differences between traditional satellites and the Starlink constellation satellites and how it’s different in terms of your relationship to the satellite itself, but one of the things that we were looking at earlier was an example where there was satellite connectivity between two stations over a body of water. So there was a body of water between them, and just how weather-sensitive this method of transport is because we could see that there was significant attenuation of the signal on this regular basis. So it was something like every seven hours because of the tide going in and out. So we look here, we can see, for example, you had this instance here where, so on the two points on the far left and the far right are basically where the satellite stations were. So the satellite dish, is that right?

Mike Hicks:
Yeah, that’s right. The two endpoints across from there, just to get to that, there’s a reason, and we have to do this situation because there’s no other way to get there. There are no submarine cables. There’s no other way, so it’s actually not a long distance. It seems ridiculous to send it up into the air and back again, but it’s just the only way to actually achieve that connectivity. So essentially, ground stations at either end of a connection.

Angelique Medina:
Mm Hmm. And so then, what we’re seeing is that on the top, there’s this, you can see these little pyramids here, and they’re representative of these, and they’re spaced the same length of time apart, and we’ll get this 9% packet loss on the regular, and then it will just go away.

Mike Hicks:
Exactly, and you can tie those directly in with the title patent, and there’s other aspects that come in because this is effectively or is coming across the body of water. And as you mentioned, that the attenuation pulls down. It has its impact on the radio waves, which is essentially what we’re talking about here. And then, even if we start to get choppy water, it can start to change these things. So you can have different patterns that can come across from there. And so, you can lose signal during the winter months where the weather’s really choppy, you’ve got a two-meter swell occurring because starts have a dramatic impact, of course, from those. So what they do in those instances, they say, you’ve got this new regular pattern. You can predict around it. So you just build the applications to sort that. You effectively start to have almost a little CDN where a lot of the information sits on site, and we just go out for updates. You can sort of shift those patterns. So the application is suited to meet the network’s requirements across from there, which is a lot of what Ray talks about when we start to get into the satellite communications.

Angelique Medina:
Yeah, yeah, absolutely. It’s really down to the application provider to tune their application in such a way so that it’s optimized for transport over something like SatComms for example. So not really something that’s necessarily controllable by the operator of the satellites than some. So that’s an interesting bit about it because of course, you can claim low latency, but from the standpoint of how is the application performing. To a large degree, that might be somewhat out of your control unless you’re of course working with application providers, and they’re incentivized to optimize their application to go over this type of communication network.

Mike Hicks:
Exactly. I mean, what you have to do is, whereas this might be part of my communications, so I need to understand what could be the characteristics of this. And exactly as you say, adapt my application to suit that. Because ultimately, as much as we want to love to think as a network is the key part, the end-user experience is the one that’s going to be the key driver in all of this and exactly right. The incense size to actually use that from an application delivery point of view, what’s in it for me to deliver that?

Angelique Medina:
That’s right. So why don’t you tell us what’s up next with your interview with Ray?

Mike Hicks:
Yeah, thanks. So I sat down with Ray Hunter. Ray Hunter is a network solutions consultant at Globis based out of the Netherlands. He’s got over 20 years experience in networking staff, and as Angela mentioned, I worked with him at the European Space Agency, where we played around with a lot of SatComms to try and optimize, deliver the application to control things like payloads on the shuttles, as well as delivering applications to scientists around the world, that come across from there. Very interesting chap, both English natives. So you might actually need subtitles underneath it, but no, it’s a very good.

Angelique Medina:
Great, okay, well, let’s take a listen.

Mike Hicks:
Well, I’m joined by a Ray Hunter. Ray is a network consultant with Globis, and he’s got a lot of experience in terms of building networks and then specifically in looking at SatComms. So what we want to talk to you today, Ray, about the SatComms, some of the new emerging technologies and just really to understand how they’re going to fit in, in terms of living applications.

Ray Hunter:
Oh, hello.

Mike Hicks:
So let’s start off. So satellite networks really come in all manner of shapes and sizes. It’s a multitude of techniques and technologies used to transmit the information. What do you consider a major driver for the usage?

Ray Hunter:
Up until now, the SatComms have been relatively limited in niche cases where we haven’t had any coverage from traditional wired or wireless Internet services. And the go-to model we’ve had has been to use a secure overlay over an insecure underlay transport network, so we would be using a standard Internet over copper, or over fiber, and then the SatComm area would be used to for the niche case where there wasn’t any coverage at all, but due to new technologies, such as SD-WAN, there’s the opportunity to combine these insecure links and these variable availability links in a manner that allows you to create an end-to-end solution of other path that’s more reliable than the individual links themselves. So I think that’s where we’re going to go is a smart path selection is going to help us to combine a lot of this technology into a single solution.

Mike Hicks:
So, where we were talking about these hard to reach places, we’re talking really geographically dispersed. Is it a competing technology of a 4G or 5G type of aspect, or there is no coverage there, and this is the only way to get data in from a SatComms perspective?

Ray Hunter:
I think it’s very dependent on the actual geography we’re talking about because in some areas, there just isn’t any 4G or 5G, and we’re also seeing that from the political perspective, there’s actually some consumer pushback for 5G because as you know, the higher frequency you use in the radio, the lower the reach. So you actually need more towers for similar coverage, and there is pushback, “not in my backyard,” for the amount of towers that are getting deployed. So there is opportunity there depending on how the pricing and the availability of these competing technologies roll out for other solutions. The traditional would be where you have a very diverse population where it’s actually too expensive to roll out the number of towers required to cover 4G or 5G. If you look at the United States compared to Europe, in Europe, you have very, very dense coverage for Wifi, and also for 3G, 4G and 5G coming up. Whereas, the pricing of these technologies in the US is really an order of magnitude higher.

Mike Hicks:
Yeah. Okay. So price obviously becomes a consideration from there as well as the coverage type of stuff. If we’re talking around some of the new systems we saw, the satellites. We touched on the LEO stuff, the low earth orbital satellites around from there. Did they give any benefits over that we would have from a traditional—when we’re thinking back in time and when you and I were talking, it was a lot of geostationary types of satellites. Now, we’re starting to see these distribution of a Low Earth Orbital stuff from there. Is there a difference in any of the perception of those systems or the expected outcome from those?

Ray Hunter:
You’re definitely battling a perception issue from history. In the geostationary situation, you had two very big problems was the latency because the salt lakes are physically far away and the speed of light restrictions effects fixed latency. And the other one was asymmetry. Was that because of the way the bandwidths were licensed and the size of the dish you are allowed to deploy, the bandwidth in one direction was very different to the bandwidth available in the other direction. And that limited their use in some applications because it assumed that you were broadcasting data from a central point to multiple endpoints. Whereas, a lot of people are actually working precisely the other way around. They are very dispersed sources of data like telemetry that they want to concentrate in a central area, and the other thing about the geostationary of course, is that you’ve only got one satellite typically. And when you hit the Equinox, the sun goes in front of that dish, and you lose your connection for several minutes. I’ve certainly come across that situation on an oil platform where someone starts shouting at you and say, “Fix the link.” I say, “Give me 10 minutes.” You come back, and the link’s working and you’re a genius because you fixed it because the sun’s moved.

Mike Hicks:
That’s excellent, so just to qualify that, I guess the geostationary, they sit around 35,000 kilometers and then the LEO’s sitting around, anything from 200 to 1200 kilometers.

Ray Hunter:
Yeah, and that’s a huge difference in terms of the speed alone.

Mike Hicks:
Yeah, absolutely. And then, the other difference, as you said, the fact from a geostationary, we’re looking at a single satellite that goes with the Earth’s orbit, and then the LEO stuff is a constellation. So we’ve got lots of little, they’re probably all the same size. We have a mesh of satellites around from there, but they essentially stay in place, and them the Earth turns. And then, we do handoffs between each one.

Ray Hunter:
And that’s another good point that says that the latency for LEO is understood, and that it’s going to be a lot less, so the path may be shorter, but I don’t think what’s very well understood is how the handoffs going to work when you’ve got multiple satellites in view, and where the constellations are incomplete because these guys are still launching the constellations or where users are located at high latitudes where the satellites may not pass as frequently overhead.

Mike Hicks:
Right, okay. I think they get impacted by where the conditions are, so space weather and whatever up from those that impact it. What about down on a Terra Firma type of position, is there considerations around as well?

Ray Hunter:
Oh, certainly. I’ve come across simple microwave links where it rains and the path bends and the microwave link no longer works over a very long hop, and you have similar of these frequencies, you have similar effects. I’ve also worked on systems that rely on the weather using tropospheric scatter to bend the radio with clouds. So this absolutely has an effect. And similarly, you have very, very physical effects that someone puts a crane in front of your dish, and well, of course, someone can dig up your fiber, but you also have line of sight issues as well with issues.

Mike Hicks:
Yeah, I had a similar one with the microwave. There was a tidal basin, and it was, as the tide dropped, it dragged the signal down, so I fully understand that, so yeah, interesting. So we’ve got, obviously, these idiosyncrasies or these conditions that we have to take in terms of the actual satellite delivery itself, but there’s obviously things you can do that treat that from a delivery perspective. What would you recommend or what do you think or when you’re tackling a satellite network or network enclosed satellite coverage, what do you do in terms of tuning that from a performance perspective to get the most out of it. We also fixed things such as the latency, or in some cases, the LEO, unknown latency, but is there anything else you can do or recommend when you start to tune this from a delivery perspective?

Ray Hunter:
I think you have to look very much at the data you’re transmitting. So there’s definitely potential for exploiting synergies with upcoming technologies like SD-WAN, where the technology points between number of differing paths with differing statistics. And then, it chooses the correct path for that particular piece of data based on how the links are behaving at that moment. So in traditional routing networks, you either had the link was up or the link was down, and then you routed based on that situation network. Now, we’re looking much more at how the path is selected based on the needs of the data itself. So you’ve got upcoming solutions, which combines a number of IPv6 connections together to form a tunnel. And you have other situation, where the programmer, instead of having to worry about all the individual links, just talks to the tunnel.

Ray Hunter:
But I think there’s a lot of scope for improvement and a new technology there because the programmers themselves are not used to dealing with this complexity. Programmers generally deal with the socket interface, which hasn’t changed for many, many years. And that does not yet reflect these new technologies of the underlying links. And also the programmers themselves are just not used to dealing with the problems that are below. So even one optimizer that can do things like local TCP arc and have application-aware, handoff to avoid the latency issues. I think there’s still a lot of development to be done there.

Mike Hicks:
So it’s a combination we’re looking at. So we’re saying the developers essentially are looking at, like you said, just consider the next hop to be the interface coming out from the server or the application from that perspective. But if they can combine that intelligence with stuff we’ve got, you mentioned the mobile IPv6 and the Shim stuff as well. Those are obviously intelligence or intelligence within the network itself. So you’re saying it’s a combination of the two to give the best delivery mechanism.

Ray Hunter:
Yeah. So there’s a very much an ongoing debate of whether we need an intelligent network or whether we need intelligent end devices. So the experience of the intelligent network has been very much colored, if you like, by the experience of X25 and those technologies from the past, which were incredibly reliable, you put your packet in there, and it came out at the other end and no questions asked, but the price wasn’t right. And now, we have a shift more towards with the experience of mobile phones whereby the network is considered dumb and the end device has all of these different radios available to it. It does wireless, sometimes wired. It has 3G, 4G, 5G, whatever you want, potentially satellite as well in the future. And then the device itself has to work out the path, and I think we’re going to come to some middle ground where there will be some technologies emerge that allow the network to provide some smart solutions, but there will also be some requirement on the end devices themselves on path selection.

Mike Hicks:
Cool. That’s interesting. So this combination, like you say, don’t think about it with a mobile phone, but you can walk from one area to other. It just picks a handoff up and changes. If I’m in the office here or the house, it switches to the Wifi. If I go outside, it picks up 4G seamlessly without me knowing, but it’s a combination you’re saying. You say that in the future.

Ray Hunter:
Absolutely, and it’s very application dependent because, to be honest, if my mobile phone doesn’t work overnight for eight hours and no one’s trying to call me, it’s not a problem, but if it drops for one second and my call drops when I’m on a call, sales call with a really important person, then I care about that. So it’s going to be very application dependent of what we’re trying to solve in terms of the solution where we deploy will be very linked to the use case that we’re trying to solve at the end of the day.

Mike Hicks:
Yeah, that’s interesting. So that brings us back, I guess, to the satellites because a lot of these new vendors who are starting to push the LEO stuff out there, from a marketing perspective, they’re talking about low latency. Essentially, what you’re saying there is that latency isn’t the thing. Well, it is a thing. It’s very important, but it’s going to depend on the application, the use case and what it is we want to achieve as a user.

Ray Hunter:
Absolutely, and they’re marketing this latency because that’s something they can do compared to geostationary. And they’ve got to win that battle of perception as you stated earlier, that they need to remove this idea that SatComms is high latency. So I can understand why they’re doing that, but it’s not from a user perspective. It’s not the be-all and the end-all.

Mike Hicks:
Yeah, absolutely. So essentially what we’re saying is we’ve got to work with every technology. We’ve got to be able to do what’s best for my location on top of my application, and what it is I want to achieve. I’m talking to a CEO, I need my call to go through from talking to you and me chatting. It doesn’t really matter so much. So, yeah. Okay. So that’s very interesting.

Ray Hunter:
And that will also change over time because the devices have become much more mobile, and it used to be that devices were just plugged in the wall, and that was it. But now the mobile device is moving with you all over the place. We’ve got wearable tech coming up. We’ve got Internet things which are placed in vineyards to monitor crops. So the Internet really is exploding and developing at this time. It’s certainly a very dynamic time in the industry.

Mike Hicks:
Yeah. And then, on top of that, I guess in those, back to about the intelligent path selection aspect, those are going to be dynamic and changing all the time, depending on what’s going on. So it’s important to understand those parts through there, and then does the characteristics match this, so yeah, I can see this need for this merging of minds as it were, the intelligence in the application or the end device combined with my networks to do almost intent-based networking to understand what’s happening in delivery and a quick path.

Ray Hunter:
Yep. Absolutely.

Mike Hicks:
Cool. Well, thanks very much, Ray, that’s been a really interesting. It’s great to talk to you, and thanks very much for your time.

Ray Hunter:
No, you’re welcome.

Angelique Medina:
That was a great interview, really interesting topic. What stood out to you in terms of having had a chance to reflect back on your discussion with him? Was there anything that we didn’t discuss earlier that was either surprising to you or you thought was particularly interesting?

Mike Hicks:
Yeah, I thought what was an interesting point was if we go back to the history of the satellites. They were put in because that was all we could get connectivity to. In fact, when I first came to Australia, the only way I could get connectivity into my house was for a satellite connectivity, but we’ve changed that over the age, and they’ve not dropped out of fashion, but just liked around there, but with the advancement of these Low Earth Orbital stuff, we’re starting to see this change now where it’s becoming, yes, it’s the best option for bang-for-buck, but it is also changing the usage of the satellites, where once they were expensive, and they were this last resort, they’re now starting to be thought of more and more, and if you take that rural type of scenario, where we’ve got, you mentioned around the FCC funding aspect from there, there’s no need to actually go and put sort of a whole bunch of 5G towers up there.

Mike Hicks:
It becomes cost-prohibitive. So this low, the LEO stuff, becomes really useful, but it also then we start to consider things like the Internet of Things and this mobile connectivity. This is the best way to connect these systems up. So I guess an interesting point for me was just moving into the mainstream, and the application owners themselves, and as well as the onus on them to deliver over it, has to be aware that these are going to be part of their delivery system and have to understand the characteristics of how the applications are going to perform across those types of links because it may well be their last-mile solution they need to satisfy. And as we said, meet that end-user experience.

Angelique Medina:
Yeah, that’s a really great point. So maybe something they need to get ahead of now.

Mike Hicks:
Absolutely.

Angelique Medina:
All right. Well, that’s our show. Thanks for joining us this week, Mike, don’t forget to subscribe. And for those of you who do subscribe, we have a free t-shirt that can pick up, just simply send a note to InternetReport@thousandeyes.com, and you can also drop in comments or suggestions or questions to that email address as well. So take care, and you can also find us and some of the videos from our event that we had a couple of weeks ago at our YouTube channel. So until next week, have a great week.

Mike Hicks:
Thanks very much.

Subscribe to the Internet and Cloud Intelligence Blog!
Subscribe
Back to ThousandEyes Blog