SpaceX Plans To Start Offering Starlink Broadband Services In 2020 (theverge.com) 125
SpaceX president Gwynne Shotwell said the goal is to complete six to eight Starlink launches to get sufficient coverage to start offering the service to consumers in 2020. SpaceNews reports: SpaceX is confident it can start offering broadband service in the United States via its Starlink constellation in mid-2020, the company's president and chief operating officer Gwynne Shotwell said Oct. 22. Getting there will require the company to launch six to eight batches of satellites, Shotwell told reporters during a media roundtable. SpaceX also has to finish the design and engineering of the user terminals, which is not a minor challenge, Shotwell acknowledged.
SpaceX CEO Elon Musk has a Starlink terminal at his house and he used it to send a tweet early on Oct. 22. "Sending this tweet through space via Starlink satellite," he tweeted to his 29 million followers. "Whoa, it worked!!" Shotwell said SpaceX will need to complete six to eight Starlink launches -- including the one that already took place in May -- to ensure continuous service in upper and lower latitude bands. "We need 24 launches to get global coverage," she said. "Every launch after that gives you more capacity." SpaceX wants to offer the service to the U.S. government but is now focused on how it will serve the consumer market. Many of the details of how the service will be rolled out remain to be worked out, she said. When possible it will be offered directly to consumers following Musk's Tesla model for selling cars. In many countries the company will be required to partner with local telecom firms to offer the service. Last week, the company requested the International Telecommunication Union to approve spectrum for 30,000 Starlink satellites that would be in addition to the 12,000 already approved by the U.S. FCC.
SpaceX CEO Elon Musk has a Starlink terminal at his house and he used it to send a tweet early on Oct. 22. "Sending this tweet through space via Starlink satellite," he tweeted to his 29 million followers. "Whoa, it worked!!" Shotwell said SpaceX will need to complete six to eight Starlink launches -- including the one that already took place in May -- to ensure continuous service in upper and lower latitude bands. "We need 24 launches to get global coverage," she said. "Every launch after that gives you more capacity." SpaceX wants to offer the service to the U.S. government but is now focused on how it will serve the consumer market. Many of the details of how the service will be rolled out remain to be worked out, she said. When possible it will be offered directly to consumers following Musk's Tesla model for selling cars. In many countries the company will be required to partner with local telecom firms to offer the service. Last week, the company requested the International Telecommunication Union to approve spectrum for 30,000 Starlink satellites that would be in addition to the 12,000 already approved by the U.S. FCC.
30K satellites (Score:4, Insightful)
The sky is quickly becoming a very busy place.
Re: (Score:3, Informative)
The sky is also a very Very VERY big place (even if there are occasional instances where orbits intersect, they use very big prediction boxes on the order of a cubic kilometer to avoid potential collisions).
Re: (Score:2)
I think that they're worried about the effect on the night sky, because of the pictures of the early launches. But those were people deliberately trying to spot them. And even a number of people deliberately trying to spot them, and knowing right where to look, failed. SpaceX has since said that subsequent versions are going to be modified to be less visible than the earlier versions.
That said, of course they'll have an effect on ground-based telescopes, for which even tiny satellites stand out like sore t
Re: (Score:2)
Airplanes are pretty common already. If you look up at night, you can actually see them even without a telescope and without looking up a specific time or part of the sky.
Ground-based telescopes also have to deal with atmospheric distortions, clouds, birds, mosquitoes and so forth. Dealing with a complicated sky is an inherent part of running a ground-based telescope.
Re: (Score:2)
The difference is that airplanes are in the Earth's shadow. Suborbital flights frequently won't be.
During their boost phase, this effect can be spectacular enough to cause car accidents [youtu.be] - although one presumes that people would get used to it.
Re: (Score:2)
How much fuel does each launch use?
Re: (Score:3)
Why is nobody worried about the pollution caused by all the (un)necessary rocket launches associated to this project?!
There's about 440 tons of kerosene in a Falcon Heavy. If we assume 150 satellites/launch, then they'd need 88000 tons. A 747 jet uses about 10 tons/hour, so that's about the same as a single 747 flying for a year.
Re: (Score:3)
They'll be switching to Starship. 400 satellites per launch for 3300 + 1200 = 4500 tonnes of propellant. Propellant is burned at a ratio of 3,8 O2 to 1 CH4, so 937,5 tonnes of CH4, or 2,34 tonnes per satellite. That's 10,2 tonnes of CO2 per satellite. For comparison, a 8l/100km car driven 20000km/yr burns 2500l of gasoline. At 2,32kg CO2/l, that's 5,8 tonnes per year. So each satellite has the embodied launch CO2 of 1 3/4 years of a typical car being driven. Putting the whole constellation up will be the
Re: (Score:2)
Re: (Score:2)
"The sky is also a very Very VERY big place...."
Some of us are old enough to remember reading that same argument about rivers, countries, continents, oceans, etc., to justify doing nothing about things like pollution. History proves mankind has a talent for filling up any available space at an alarming rate.
Re: (Score:2)
Exactly what I was thinking. Yes, the sky is big... until you realize you have to keep track of all those many, many objects, large and small, moving at very, very high speeds for many, many years. One accidental collision and each of the objects involved would turn into thousands of untraceable small projectiles moving in all directions at very, very high speeds.
Maybe this is the Great Filter: surround your planet with debris in such a way you would never be able to escape it.
Re: (Score:3)
This is a misconception. Tiny bits of debris - the stuff that's difficult to track - does not stay in orbit for long ("many, many years"), when you're talking about LEO collisions - particularly for low-altitude satellites, which represent the lion's share of the Starlink constellation. And the longer it's in orbit, the greater the fraction of its time it tends to spend in areas where it doesn't intersect other satellites.
Remember that kinetic energy is proportional to mass is proportional to volume is pr
Re: (Score:2)
Remember that kinetic energy is proportional to mass is proportional to volume is proportional to radius squared - but drag is proportional to cross section is proportional to radius squared. So the smaller the object, the faster it deorbits.
I agree but one detail is wrong.
Mass is proportional to radius *cubed*. Cross section is proportional to radius squared. So as the radius increases, the mass increases faster than the cross section so the energy (to deorbit) increases faster than the drag.
Re: (Score:2)
At our current rate of population growth, humanity will have consumed all mass in the observable universe and turned it into humans in only 10,000 years. We know that can't happen, because of the speed of light and all that, so we will hit a wall.
With some difficulty resources were diverted to permit the construction of a small but important device. It was a time machine. With one volunteer aboard (selected from the 900 trillion who applied) it went back to the year 1. Its cargo was only a hunting rifle with one cartridge, and with that cartridge the volunteer assassinated Snodgrass as he trudged up the Palatine.
Re: (Score:2)
So will the ground if this becomes popular. You need a "pizza box size receiver" according to Musk and it must be able to see the sky, which means it will probably have to be mounted outside. A satellite dish can be on the side of a building because the satellites are in geosync orbits, but the SpaceX ones are not so presumably the receiver can't be partially blocked by a wall if you want 24/7 coverage.
Not sure what you do if you live in an apartment.
Re: (Score:2)
If you live in an apartment, you buy the 5G service which is sold to you by a company that owns a "pizza box" in the neighborhood.
Re: (Score:3)
Adding a pizza box sized receiver to our bus is going to be easy, so that's ideal for me. If you live in an apartment, you probably don't need Starlink because you can probably get cable. That will have less latency, so even though the LEO latency is much better than what us GEO users are having to put up with now, cable is still more desirable.
For boondockers and RVers Starlink is going to be a godsend, because right now people are using cellular with signal boosters. There are literally zero companies sel
Mobile use (Score:2)
Adding a pizza box sized receiver to our bus is going to be easy,
That's the kind of use I immediately though about: people hiking/camping/traveling away from civilisation - it's going to be easy to mount one on the vehicle.
If you live in an apartment, you probably don't need Starlink because you can probably get cable.
Or in several European cities: get several direct fiber providers, sometimes even directly connected to your appartment.
You can get self-aiming for satellite TV, but not for internet.
Is there a GEO sattelite up there that happens both carries signal for TV and signal for some internet service provider ?
Re: (Score:2)
Is there a GEO sattelite up there that happens both carries signal for TV and signal for some internet service provider?
I don't think so. ViaSat sells both satellite TV and satellite internet, but I believe they're on separate satellites. I'm on ViaSat-1 [wikipedia.org]...
Re: (Score:2)
$100 says they'll start building starlink antennas into Teslas within 2-3 years.
Re: (Score:2)
Starlink will have less latency than cable. c is 66% pushing through glass due to refraction. With satlink there is no glass to push through
Re: (Score:2)
Mounting the receiver on an external wall or balcony should work just fine. You want to point it close to straight up because that's the shortest distance to the nearest satellite.
The reason they need so many is so you don't have to point it in a particular place (like a geosynchronous satellite) or have it track.
Re: (Score:2)
They describe it as a phased array which suggests that it does in fact track.
Re: (Score:2)
Yeah, they have to be able to track a bit. I meant that they don't have to track a single satellite across the whole sky.
GPS uses 32 satellites, of which you need to see at least four, and there's no problem getting a GPS fix except if you're really hemmed in by buildings. GPS signals are lower frequency so they'll diffract a bit more.
Starlink is going to be thousands of satellites, and you only need to see one at a time, so you should be fine even with a quite restricted view of the sky.
Re: 30K satellites (Score:3, Interesting)
Re: 30K satellites (Score:5, Insightful)
I will be following Starlink with exceptional interest.
Enticing (Score:2)
Broad band, broad latency.
Re: (Score:3)
This has been covered right here on Slashdot many times; the low orbit insure that you can have just as low latency if not less than with wired connections on average if you route packets optimally. Especially if you route between satellites.
Re: (Score:2)
Thanks, that surprises me, will look into it.
Re: (Score:2)
Light moves at ~300 thousand kilometers per second
LEO satellites range from a couple hundred to a couple thousand kilometers altitude.
Do the math. :) Even accounting for round-trips and cosine losses, transmission imposes only minimal latency - and as noted, lets you skip fibre delays and often the numerous routing delays that make up the last-mile problem. Latency in the system is generally expected to be superb.
GEO is where you get latency problems - ~36 thousand km altitude. But Starlink is a LEO cons
Re: (Score:2)
RTT to the satellite and back to the ground station will be 35-45ms, but that's not the only source of latency. Each satellite shares all available bandwidth between clients using time slicing, you you have to wait for your time slot to come around before transmitting. Then you have the normal latency from the ground station to the server.
It's not too bad compared to ADSL or older cable installations. Compared to fibre it sucks though.
It will be interesting to see how well it works for stuff like gaming.
Re: Enticing (Score:2)
> Compared to fibre it sucks though.
No, RTT, for long distances, is shorter than fiber because light propogation through fiber is much slower than through space. So much so that an early idea people had for a target audience was international high-frequency trading platforms. At gigabit speeds medium latency is minimal.
Re: (Score:2)
But how often is international latency that important? Most of the time the server is near you, part of a CDN, so the ground route is faster even with 30-40% slower propagation in fibre.
With FTTC you normally see pings for common services like Google or XBOX Live under 5ms, which is about 1/10th the time it takes just for the satellite signal to go up and back down, let alone get routed to the server. In Japan I was seeing 1ms ping times for Google.
Lietteraly opposed groups of customers (Score:2)
Compared to fibre it sucks though.
Except that Starlinks targets the exact extreme opposite end of the range of users.
- Fibre is mostly aimed at densely populated area, where it's nearly trivial to link a single appartement building and thus dozens or more of user, sometime even directly reaching fiber inside their appartement (a.k.a. a normal regular situation in many European large cities). Also 4G/5G fits a similar use pattern except for mobile users.
- Starlink is mostly aimed at mobile users in the middle of nowhere (think users mounting
Re: (Score:2)
Re: (Score:2)
in rough terms, millisecond and a half to two milliseconds.
hawk
Why is a Starlink terminal needed? (Score:2)
Re:Why is a Starlink terminal needed? (Score:5, Informative)
It's just the name for the router/antenna that you place in your house. It's the point where the satellite connection terminates.
Re: (Score:2)
What's wrong with using a term anyone can understand? Does the FUD magic cease to work if he did?
Re: Why is a Starlink terminal needed? (Score:2)
Do you have similar issues with terminal illness, and a computer terminal?
Terminal means different things to different contexts.
Get used to it.
Re: (Score:2)
What's wrong with using a term anyone can understand?
Name a computing term that anyone can understand. By the way is it safe to shutdown windows by pressing the power button on the harddrive?
Re: (Score:2)
Someone built a computer you can just turn off.
The idiocy of modern computers is software engineers cannot make the quantum conceptual leap necessary to stop bubbling the filesystem API up to something the user is forced to deal with.
Re: (Score:2)
No thanks. I'd prefer software to be able to do its thing and shutdown gracefully as I walk away from the machine. Far better than having it waste my time while I'm sitting there, or worse, when it's starting.
Also dealing with filesystem closure at shutdown is precisely the opposite of forcing the user to deal with it. Software engineers dealt with that concept a long time ago, no idea why you complicate it with quantum gobbeldegook.
Re: (Score:3)
Satellite terminals are called satellite terminals. It's not something Musk made up.
But sure, we should dumb everything down so the lowest common denominator doesn't have to potentially learn something. Like calling cable modems "routers".
Re: (Score:2)
Re: (Score:2)
Starlink Terminal is just a rebranded Minitel terminal, but instead of plugging it into the phone jack in your house, you use a satellite dish.
Okay, I admit I have no idea what I'm talking about.
Re: (Score:2)
It's a phased-array receiver about the size of a large pizza box. The exact design hasn't been unveiled, but you can expect it to be pretty much like any home gateway in terms of providing wifi, ethernet ports, etc.
Re: (Score:2)
The exact design hasn't been unveiled, but you can expect it to be pretty much like any home gateway in terms of providing wifi, ethernet ports, etc.
Well, the question is whether it's going to be one- or two-piece. If the whole thing is contained in the box then it's not unlikely that it will have just one PoE port, and come with just a couple ethernet cables and an injector. This is how WISPs tend to operate, IME. Then you can use your own WiFi router, and Starlink will probably offer to rent and/or sell you one.
Exede doesn't do this. The box is in the house and gets RF from the dish outside. Presumably they want to be able to replace the box without h
Must-buy for high frequency traders (Score:2)
If StarLink is capable of achieving lower latency than conventional routes between key locations like New York and Chicago or New York and London, it'll immediately become something that every high-frequency trader must buy, at nearly any cost. That's how Musk is going to make back his investment.
Swings in commodities prices on the Chicago Mercantile Exchange reliably impact prices on the New York Stock Exchange. If you're in New York and know what's happening in Chicago even a millisecond before everyone
Re: (Score:2)
yeah. in fact, this is yet another proof that we have been contacted by aliens... the Ferengis !
Speaking of latency (Score:2)
Re: (Score:3)
Re: (Score:2)
HFT isn't useless. It's used for arbitrage, which means that regular stock traders get to benefit from synchronized prices around the world, giving them access to a bigger market for a tiny fee.
Re: (Score:2)
> It would be sad if something as useless as HFT is the business case for StarLink.
Starlink has a business case for HFT, but HFT isn't a business case for Starlink.
Starlink's purpose is to fund Mars exploration. Required revenues are in the tens to hundreds of billions of dollars a year. Its method is to provide high-speed Internet access for everybody on Earth.
There simply isn't enough money among the HFT traders to pay that much, no matter how valuable the service is. HFT people may be a customer l
Re: (Score:2)
I thought about that too, but apparently high-frequency traders now have dedicated microwave links for critical data.
Re: (Score:2)
It's not so much the distance but rather the coding and processing delays.
Of course, wired connections eliminate de facto any coding and processing delays.
Re: (Score:2, Informative)
Wired connections have those as well, obviously, but there will be less, as an optic fiber is more reliable and constant. And you'll probably need at least 4 satellites to make a connection between New York and London, adding more processing delays.
Extra distance isn't trivial either. Satellites will be orbiting at 550 km, so that's at least an extra 1100 km one-way latency, which is 20% of the distance between New York and London.
Re: (Score:3)
Satellites will be orbiting at 550 km, so that's at least an extra 1100 km one-way latency, which is 20% of the distance between New York and London.
And, speed of light in an optic cable is only 66% of the speed of the signal between satellites and Earth bases. So even with the 20% detour you mention, satellites are still winning.
Re: (Score:2)
Guess that 20% still makes it a near tie then, as light travels about 30% slower through fiber than electrons through free-space.
https://physics.stackexchange.... [stackexchange.com]
Re: (Score:2)
FTFY
Re: (Score:3)
It's unlikely to be lower latency than an optic fiber across the ocean.
No. Speed of light in vacuum is 40-50% greater than in glass fibre. That is big!
Re: (Score:2)
Aahhh.. That's a good point. I read that it's 31% in normal silica glass, though, so it will be a close race.
Re: (Score:2)
Aahhh.. That's a good point. I read that it's 31% in normal silica glass,
Maths : "31% slower" means 69% of the speed. Reciprocal is 1/0.69 = 1.45, so vacuum is 45% faster. Same as glass being 31% slower.
Re: (Score:2)
I'd rather be 45% longer than someone else, than that they be 31% smaller.
Re: (Score:2)
https://www.extremetech.com/co... [extremetech.com]
Re: (Score:2)
https://www.extremetech.com/co... [extremetech.com]
Interesting, but will not be replacing long-distance cables due to the 3.5 dB/km losses.
If they solve that, the Starlink will lose a lot of money.
Re: (Score:2)
Re: (Score:2)
High speed connections for stock trading have already minimized routers between endpoints.
Re: (Score:2)
Yeah, but the biggest win will come from minimizing the hops. Terrestrial networks have a lot of routers and repeaters between you and most hosts you want to talk to.
-jcr
I suspect the low-latency traders are already using leased fibres and super-fast nodes, not off-the-shelf IP routers and internet.
But from here, ping times to US are around 200ms and 70-80% of light-speed in fibre. So light-speed is substantial.
Re: (Score:3)
30ms for one-way is already available:
http://www.qpdc.co.uk/the-equi... [qpdc.co.uk]
Re: (Score:2)
30ms for one-way is already available:
http://www.qpdc.co.uk/the-equi... [qpdc.co.uk]
But that's only 5567km by surface. Would be 18.5ms in a vacuum.
Re: (Score:2)
But also the distance is further due to taking a less direct route. Particularly at first the spacing of ground stations will be an issue too.
CDNs exist to mitigate a lot of this anyway. And satellite will never compete with fibre where you can get a 1-2ms ping times to popular services. The first time I pinged google.com on 2 gigabit fibre and it said 1ms I thought it was an error.
Re: (Score:2)
The first time I pinged google.com on 2 gigabit fibre and it said 1ms I thought it was an error.
You were not actually pinging google's data farm, but a nearby caching node. Possibly a google box at your local ISP. Must have been less than 100km away.
Re: (Score:2)
Yes, that's the point. They all use CDNs and most ISPs have caches for popular services now.
Even ones that can't be cached like game servers see sub 5ms pings on fibre in the UK, and the UK's infrastructure is pretty poor.
Re: (Score:2)
These guys [qpdc.co.uk] are bragging about a 30ms one-way trip, which seems to be state of the art. Putting the distance at 5600km, plus ~2500km for the altitude of the satellites, we'd have a travel time of ~27ms for light in vacuum. That gives you ~2ms to handle processing delays and path imperfections. That might be feasible? I've also put pretty conservative numbers to my estimates.
I think you'd really see the benefits in overland connections where geography makes it impractical to lay a straight cable. Or in lo
Re: (Score:2)
Satellites only operate at 550 km altitude, so that's 1100 km one-way. But at 550 km, the line of sight distance over the horizon is 1600 km, which means you need two intermediate hops to cover New York to London. And those two hops are not going to be in a straight line, adding a bit more distance.
...required to partner with local telecom firms... (Score:2)
Right off the bat Canada is fucked.
Re:...required to partner with local telecom firms (Score:4, Informative)
Well, it has to get into and out of the local ground-based internet somehow, with as few of hops as possible. You don't want all Canadian communications to have to hop via Starlink over to the US and then back to Canada every time you want to look up your local news, do you? :)
The ground stations are designed to connect straight into backbone links. Starlink is intended to solve the "last link" problem, functioning primarily as an ISP, rather than to take over the role of an internet backbone. That said, it does have intra-network relaying capacity, and that can only be expected to grow with time.
Re: (Score:2)
Oh, no, it's meant to be a backbone. The speed of light in space is 2x faster than in glass (fiber cable) so the trans Atlantic link on Starlink is going to have far lower latency than using the existing fiber channels across it. This is supposed to be a big win for high speed trader in th
Re: (Score:2)
Since when do satellites use light to communicate?
Re: (Score:2)
Since they use lasers, of course.
https://en.wikipedia.org/wiki/... [wikipedia.org]
Re: (Score:2)
so much faster than string, and you don't have to hang around the kitchen waiting for mom to open a can you can snag after she empties it . . . :)
hawk
Re: (Score:2)
Since Sputnik?
Photons (Score:2)
Satellite to sattelite links are supposed to be done using litterally light - a.k.a. 'Fricking lasers !' [wikipedia.org] (Sharks provided separately).
Satellite to ground is classically done using microwaves. Not what is colloquially called "light", but still electromagnetic light carried by photons, so all the phyiscal values (such as speed) still do apply.
Note that lots of optical fiber emply infra-red so again not technically visible light, but yet again still subjected to the same physics.
Re: (Score:2)
Starlink is intended to solve the "last link" problem, functioning primarily as an ISP, rather than to take over the role of an internet backbone. That said, it does have intra-network relaying capacity, and that can only be expected to grow with time.
The top major US ISPs have been taking over both roles of late.
Looking at my own ISP, twtelecom, my data starts from the "last mile" yet travels nothing but twtelecom's own network all the way into California before hopping to quest and americanis where slashdot is hosted, and both of those last two hops are within the same exchange, likely in the same data center building.
My experience with Comcast is similar.
So yes I would presume you're correct that Starlink will be intending to be a "last mile" ISP, but
Can't wait (Score:3)
I'm interested to see what this does to monopolies and duopolies all around the world not to mention places like North Korea and China will probably be scared shitless of such an endpoint making it across the border.
Re: (Score:3)
North Korea and China are pretty safe, considering the government exerted control on entry of hardware on their soil...
Re: (Score:2)
They can probably detect people using those endpoints and then pay them a visit, or operate a jammer.
Re: (Score:2)
Nah, the ground stations in China will just plug straight into the Great Firewall.
Starlink *could* be designed as a censorship avoidance tool, particularly for a small country like North Korea (it wouldn't be hard to hop all communications over to ground stations in South Korea). Who knows for a case like that**. But China? No way that they try to do it in a manner that's in violation of local laws.
** - If I had to guess, I'd guess that they'd have a policy along the lines of, "Sorry, we don't offe
Re: (Score:2)
It requires a pizza box size receiver with a view of the sky. China and NK won't have too much trouble finding them. NK might just jam 24GHz anyway, and China might decide to put up its own network.
SpaceX may have to disable service as the satellites fly over China. China doesn't have to give them a licence to use that spectrum. If they refuse then China will no doubt be happy for the opportunity to run some anti-satellite weapons tests. A laser might be enough to destroy those things since they apparently
Re: (Score:2)
So long as Musk sends the Chinese customers' data to ground stations in China, there's no reason for China to complain. They'll extract whatever amount of income they like from the ground stations, and install their monitoring equipment there. I don't think Musk is dumb enough to try to go around China.
Re: (Score:3)
Why? IntelSat already provides satellite Internet for North Korea. No corporation is going to offer services in a country that doesn't allow it. It isn't like satellite Internet is some new thing.
Re: Can't wait (Score:2)
I can imagine a day when someone sends up a Cubesat that hacks into the space internet link and hosts pirated or free speech content. Which reminds me, will Starlink support ipv6? I assume so.
say goodbye (Score:3)
Starlink will be far worse. as there will be satellites in every direction you look overwhelming the 10-13 ghz bands. And while we can excise RFI to a degree in post processing for some problems, we cannot do this for continuous interference in the frequencies above 5ghz. This could change in the future, but for now the lower frequency range of 1 to 20 gigahertz is in bad shape when it comes to collecting clean data.
Re: (Score:2)
You correctly qualified "earthbound".
SpaceX is selling a problem and selling a solution. With much lower costs to orbit, space-based astronomy will become far more practical.
I do agree that it is unfortunate that earthbound radio (and optical) astronomy will suffer, as it will always be the lowest cost and lowest barrier to entry, but I think the ability to have quality Internet all over the globe is worth it.
Cars (Score:2)
So does this service need a dish or can you hide an antenna in a car?
Could be useful as leverage when negotiating deals with cell companies for coverage for the cars.
Re: (Score:2)
My understanding is that it's 'large' (pizza box?) and requires stability. So... not for cars.
why (Score:2)
Re: (Score:2)
"Never underestimate the bandwidth of a rocket full of tapes hurtling through space"
Re: (Score:3)
What's the bandwidth of a BFR loaded with hard drives?