If any of the many plans for global, distributed satellite internet come to fruition, we're not going to be giving two fucks about those cables fifteen years from now, because they'll be old,...
If any of the many plans for global, distributed satellite internet come to fruition, we're not going to be giving two fucks about those cables fifteen years from now, because they'll be old, slow, expensive to repair, and obsolete. Also, let's not pretend it's a herculean task to run the cable to a new location. Sure, it's a pain in the ass, but hardly as bad as running the cable across the ocean in the first place. In fact, we may not need to move them at all - I seriously doubt New York City is going to pack up and move just because the water rises fifty feet. They'll build seawalls, it's cheaper than relocating a multi-trillion dollar chunk of real estate housing tens of millions of people.
Yes, Wall Street will care. The rest of the world isn't going to miss a few milliseconds, and frankly, we should put some bans on high frequency trading so that it doesn't matter to Wall Street as...
Yes, Wall Street will care. The rest of the world isn't going to miss a few milliseconds, and frankly, we should put some bans on high frequency trading so that it doesn't matter to Wall Street as well, but that's a whole other discussion/rant.
In the 90s they were calling this satellite network Teledesic. It's been around 'in theory' for 30 years. Motorola, AT&T, Verizon, everyone's looking at taking the leap. The expense of rocket launches was the chief barrier, but SpaceX has made progress there so it's not nearly as expensive as it was in the 90s. The size/cost of the satellites has come down as well, into the 'disposable' range.
Musk is going all-in with his Starlink project. It really doesn't matter who does this - someone will do it, and the bandwidth of the network beats anything, anywhere that has a 'last mile' deployment factor at play. A few thousand satellites will provide more bandwidth than the entire terrestrial internet capability at present.
Not enough to care even for the most twitchy games - 10 ms here or there doesn't make or break most games. Also once the network is mature, it'll be faster than fiber optics for long distances, so...
Not enough to care even for the most twitchy games - 10 ms here or there doesn't make or break most games. Also once the network is mature, it'll be faster than fiber optics for long distances, so that imbalance will swing the other way. It's rare to see something that can actually outclass fiber optic transmission.
The next big thing in satellite internet is massive (>1000 satellite) constellations in Low Earth Orbit. Traditional constellations are in higher orbits so that fewer satellites can be used to...
The next big thing in satellite internet is massive (>1000 satellite) constellations in Low Earth Orbit. Traditional constellations are in higher orbits so that fewer satellites can be used to serve greater areas, which leads to significant latency from speed of light delays. Low Earth Orbit is less than 1000km up, so you don't have the distance problem as bad, and the speed of light through a vacuum is nearly 50% higher than through fiber optic cable. That, and you can route transmissions without having to worry about right of way for laying cables, so in general, latency should be better on a LEO satellite constellation for connections longer than a few hundred km.
Exactly. I remember reading the protocols for transmission they were building back when it was Teledesic. It's fascinating, non-traditional router problem - a gigantic, dynamic mesh network where...
Exactly. I remember reading the protocols for transmission they were building back when it was Teledesic. It's fascinating, non-traditional router problem - a gigantic, dynamic mesh network where the endpoints are all localized to a specific group of satellites (the ones over the client at that moment, and they move, so that changes fast). The network also has to have a lot of signal redundancy so it can cope with solar flares and other more local disruptions. They have to build the mother of all collision avoidance/detection systems to handle literally billions of devices hopping on and off the network at all hours of the day. Now let's add the joy of finding a proper global signal spectrum to reserve just for this network. As I understand it, all of this pretty much done already on paper, we're just waiting on the hardware to get to space. Starlink's first test sats are up there right now, they were launched in Feb this year.
The promise of the network is for you to be standing in the middle of the Sahara desert and still have a viable 10 gigabit connection. It does get a bit tricky - the more the endpoint has to move, the slower the transmission becomes, so I expect cell towers are also going to evolve into local uplinks to this network over time.
I don't think Starlink is targeting 10Gbit for the end users just yet, there are hard limits on bandwidth with wireless tech, and with congestion you would need to be in the Sahara to get speeds...
I don't think Starlink is targeting 10Gbit for the end users just yet, there are hard limits on bandwidth with wireless tech, and with congestion you would need to be in the Sahara to get speeds approaching that. But if they could even get 15-30Mbit worldwide, that would be absolutely revolutionary.
I saw some articles claiming they could do it, though I remain skeptical given the complexity of the network and the congestion problems you mentioned.
I saw some articles claiming they could do it, though I remain skeptical given the complexity of the network and the congestion problems you mentioned.
One of the main intended use-cases of Starlink is as network backhaul. I would guess that the articles misunderstood Starlink offering 10Gbit connections to, say, cell towers, as offering them to...
One of the main intended use-cases of Starlink is as network backhaul. I would guess that the articles misunderstood Starlink offering 10Gbit connections to, say, cell towers, as offering them to end users .
The long distance routing makes it really interesting for low latency needs. It turns out that the laser based sat to sat network is where physics gives StarLink an advantage over fiber. This is...
The long distance routing makes it really interesting for low latency needs. It turns out that the laser based sat to sat network is where physics gives StarLink an advantage over fiber. This is due to the fact that IIRC, direct open laser links have a 40% speed(latency) advantage over fiber. That is because light does not travel straight down a fiber, it bounces off the internal edges of the fiber which greatly increases the distance travelled. So a StarLink route would go from your antenna directly to a satellite, then be routed via the faster space based laser network to the downlink that is nearest to your final destination. Theoretically it is faster than a fiber connection could do over the same route, given some distance.
Edit: I did some research on existing phased array antennae and found that this is the stumbling block for consumer endpoints. The price is crazy high right now, around 50k USD. I can dig up the numbers I found if anyone is interested.
Also, regarding congestion, Musk said the most they can do is around one unique endpoint per square mile in v1. This is due to the current limitations of focusing the PAA on the satellites.
This system is really begging for lots of tower links to help cover wider areas. Those towers can act like local switches and handle the larger number of connections, packaging up and unpacking...
This system is really begging for lots of tower links to help cover wider areas. Those towers can act like local switches and handle the larger number of connections, packaging up and unpacking all of the connections going up to the sat network and back down again. That lets the sats focus on one solid connection to handle a myriad of conversations. It also means we're free to experiment with last-mile tower-to-device stuff without overhauling the sats themselves, which would get pretty expensive pretty fast.
If any of the many plans for global, distributed satellite internet come to fruition, we're not going to be giving two fucks about those cables fifteen years from now, because they'll be old, slow, expensive to repair, and obsolete. Also, let's not pretend it's a herculean task to run the cable to a new location. Sure, it's a pain in the ass, but hardly as bad as running the cable across the ocean in the first place. In fact, we may not need to move them at all - I seriously doubt New York City is going to pack up and move just because the water rises fifty feet. They'll build seawalls, it's cheaper than relocating a multi-trillion dollar chunk of real estate housing tens of millions of people.
Yes, Wall Street will care. The rest of the world isn't going to miss a few milliseconds, and frankly, we should put some bans on high frequency trading so that it doesn't matter to Wall Street as well, but that's a whole other discussion/rant.
In the 90s they were calling this satellite network Teledesic. It's been around 'in theory' for 30 years. Motorola, AT&T, Verizon, everyone's looking at taking the leap. The expense of rocket launches was the chief barrier, but SpaceX has made progress there so it's not nearly as expensive as it was in the 90s. The size/cost of the satellites has come down as well, into the 'disposable' range.
Musk is going all-in with his Starlink project. It really doesn't matter who does this - someone will do it, and the bandwidth of the network beats anything, anywhere that has a 'last mile' deployment factor at play. A few thousand satellites will provide more bandwidth than the entire terrestrial internet capability at present.
Would it be enough to affect game latency?
Not enough to care even for the most twitchy games - 10 ms here or there doesn't make or break most games. Also once the network is mature, it'll be faster than fiber optics for long distances, so that imbalance will swing the other way. It's rare to see something that can actually outclass fiber optic transmission.
The next big thing in satellite internet is massive (>1000 satellite) constellations in Low Earth Orbit. Traditional constellations are in higher orbits so that fewer satellites can be used to serve greater areas, which leads to significant latency from speed of light delays. Low Earth Orbit is less than 1000km up, so you don't have the distance problem as bad, and the speed of light through a vacuum is nearly 50% higher than through fiber optic cable. That, and you can route transmissions without having to worry about right of way for laying cables, so in general, latency should be better on a LEO satellite constellation for connections longer than a few hundred km.
Exactly. I remember reading the protocols for transmission they were building back when it was Teledesic. It's fascinating, non-traditional router problem - a gigantic, dynamic mesh network where the endpoints are all localized to a specific group of satellites (the ones over the client at that moment, and they move, so that changes fast). The network also has to have a lot of signal redundancy so it can cope with solar flares and other more local disruptions. They have to build the mother of all collision avoidance/detection systems to handle literally billions of devices hopping on and off the network at all hours of the day. Now let's add the joy of finding a proper global signal spectrum to reserve just for this network. As I understand it, all of this pretty much done already on paper, we're just waiting on the hardware to get to space. Starlink's first test sats are up there right now, they were launched in Feb this year.
The promise of the network is for you to be standing in the middle of the Sahara desert and still have a viable 10 gigabit connection. It does get a bit tricky - the more the endpoint has to move, the slower the transmission becomes, so I expect cell towers are also going to evolve into local uplinks to this network over time.
I don't think Starlink is targeting 10Gbit for the end users just yet, there are hard limits on bandwidth with wireless tech, and with congestion you would need to be in the Sahara to get speeds approaching that. But if they could even get 15-30Mbit worldwide, that would be absolutely revolutionary.
I saw some articles claiming they could do it, though I remain skeptical given the complexity of the network and the congestion problems you mentioned.
One of the main intended use-cases of Starlink is as network backhaul. I would guess that the articles misunderstood Starlink offering 10Gbit connections to, say, cell towers, as offering them to end users .
The long distance routing makes it really interesting for low latency needs. It turns out that the laser based sat to sat network is where physics gives StarLink an advantage over fiber. This is due to the fact that IIRC, direct open laser links have a 40% speed(latency) advantage over fiber. That is because light does not travel straight down a fiber, it bounces off the internal edges of the fiber which greatly increases the distance travelled. So a StarLink route would go from your antenna directly to a satellite, then be routed via the faster space based laser network to the downlink that is nearest to your final destination. Theoretically it is faster than a fiber connection could do over the same route, given some distance.
Edit: I did some research on existing phased array antennae and found that this is the stumbling block for consumer endpoints. The price is crazy high right now, around 50k USD. I can dig up the numbers I found if anyone is interested.
Also, regarding congestion, Musk said the most they can do is around one unique endpoint per square mile in v1. This is due to the current limitations of focusing the PAA on the satellites.
This system is really begging for lots of tower links to help cover wider areas. Those towers can act like local switches and handle the larger number of connections, packaging up and unpacking all of the connections going up to the sat network and back down again. That lets the sats focus on one solid connection to handle a myriad of conversations. It also means we're free to experiment with last-mile tower-to-device stuff without overhauling the sats themselves, which would get pretty expensive pretty fast.
TL;DR we're all fucked.