Germany and France are introducing a government-backed project to develop European cloud infrastructure in an effort to help local providers compete with U.S. technology giants, which dominate the global cloud market.
Amazon.com Inc. and Microsoft Corp. criticized the initiative announced this week, called Gaia-X, saying the project will restrict data services along national borders.
The reach of Amazon, Microsoft and other U.S. giants worries European politicians and corporate executives. Companies in Germany and France, the continent’s economic powerhouses, and in other European Union countries are concerned about depending on technology providers that must comply with the U.S. Cloud Act, WSJ Pro Cybersecurity reported in October. The 2018 law requires American firms to provide law enforcement with customers’ personal data on request, even when the servers containing the information are abroad.
The European Union should’ve invested in efforts like this years ago, but rather late then never. And of course, it’s entirely unsurprising that US cloud providers are unhappy about this move, but that really shouldn’t be of any European legislator’s concern.
there are countries that require that user data is stored and processed in DC located in that very country, i’ve had such a problem in Russia – you may setup a cloud but on russian soil, if yo udo BI/analytics on that data and personal user data is to be stored only there.
Given that now there is GDPR i can only assume things are stricter.
Loved to read the TikTok’s news just below about a Chinese acquisition of an US based company which is causing such a “national security” concern. Fun that when it’s the other way around, US buying abroad companies, it’s no problem at all and/or nobody have to complain against a “democratic move toward freedom”.
Sounds like Galileo https://en.wikipedia.org/wiki/Galileo_(satellite_navigation)#Tension_with_the_United_States
And to combine the two – what better place for data centers then in orbit? Lots of energy and cooling.
if you die in space, your body stays warm for quite a while. it can only radiate the heat away.
so, cooling is a problem in space.
You are right on the humans but it is not a problem. Human are a worst case. We are mostly water so have very high heat capacity. A human weighing 80Kilos would take maybe 5 minutes to cool one degree (wild guess – hard to research). However this is not insubstantial energy output, i.e. cooling. Think of boiling 80 Kilos of water by cooling your PC. It would take a while. Think of boiling 80 Kilos of water using solar panels.In space energy terms humans take a lot to heat up and cool down.
lapx432,
I don’t think we can hand wave this problem away. While the lack of atmosphere limits convection heat transfer, that works in both directions. We’re no longer gaining heat energy through atmosphere, but we’re no longer loosing heat energy through atmosphere either. The principal way to transfer heat in space is radiation.
While it’s true that in the shade (or deep space) we’d eventually radiate all of our heat away, We’re actually talking about satellites that are in the presence of radiation and constantly absorbing solar energy. The average temperature at our distance from the sun is not really that cold, it’s actually similar to the temperatures here on earth, which makes sense in the grand scheme of things.
https://sciencing.com/can-astronomers-tell-distant-objects-temperature-is-9560.html
Solar panels are only about 20% efficient, the rest of the solar energy ends up heating the structure. Also, the servers themselves would overheat under their own power if you don’t do anything to take the heat away. Yes they would eventually reach a steady state temperature where they’re so hot that they naturally turn that heat into radiation (similar to an incandescent light bulb) but this temperature is likely far above the temperature normal electronics are rated to run at. Natural radiation is obviously insufficient to cool servers here on earth, it would not be different in space.
I’m not an expert on these things, but hopefully I’ve raised enough points to make the case for not hand waving the problem away 🙂
Black body radiation is works regardless of ambient temperature as it is driven by the difference in local temperature (you say 10C in orbit) to deep space (approx -270C) and not to the surrounding air. It is possible to use this radiative cooling to create ice on earth when the surrounding temperature is not below zero. See link and quote below.
In my 96 square mile calculation I assumed twice the area of black radiator to the area of solar panel. This was an informed highly conservative guess. In space solar panels are at best 30% efficient which means that by definition they have to be radiating the rest of the hear energy from the sun away. So if solar panels can radiate 70% of their heat and generate 30% as electricity then they are actually radiating 2.33 times their electrical output away as heat. You can’t have it both ways where solar panels are inefficient but don’t melt and at the same time it is not possible to radiate more than the electricity energy they generate as heat into space. That would be a physical contradiction.
Even though it is 10C in orbit, the wind chill / wind heat factor is tiny. The individual gas atoms are indeed vibrating at the same rate as 10C gas in earth, but its heat capacity (back to that again 😉 ) is minuscule so its ability to keep anything warm or cool is minuscule. This makes radiative cooling (and heating) far more effective in space. But humans still have a lot of heat by body weight with low surface area, unless they have been squashed, And their suits tend to be white…
https://www.fieldstudyoftheworld.com/persian-ice-house-how-make-ice-desert/
” The process of ice making
Even though winter nights can be very cold in the deserts of Iran, temperatures rarely drop below freezing. But ice could be made even at temperatures just above freezing, thanks to a phenomenon known as night sky radiation or radiative cooling. “
lapx432,
In a stable system, the heat input (radiation+convection) equals heat output (radiation+convection). I don’t disagree with you that it’s possible to eliminate radiation input while maintaining radiation output, and therefor become marginally cooler than the environment. However this completely ignores the energy being introduced into the system when your data center gets turned on.
I looked it up and the top 10 data centers range from 80 to 150 megawatts.
http://worldstopdatacenters.com/power/
Here on earth, most of that heat gets pumped into the environment (aka our planet). And our planet acts as a huge thermal heat sink, which will radiate the heat into space.
However if you have the same amount of energy on a satellite, but neglect to provide huge thermal radiators, you are going to end up with far less surface area to radiate thermal energy, consequently it’s going to get hot.
Any object in space WILL eventually reach a state state and output radiation energy equal to the input energy. Take a hypothetical extreme just to make the point clear, if you pumped megawatts of energy into a planet, it would barely change temperature. But if you keep pumping megawatts of energy into objects that have smaller and smaller surface areas, they will become extremely hot before reaching a steady state temperature, even exceeding the surface temperature of the sun at some point.
So while you can take advantage of thermal radiation, you still need a huge passive radiator to keep temperatures low. Cooling in space is not something you can just do away with if your servers are consuming massive amounts of electricity.
lapx432,
It’s an interesting way to avoid earth-based jurisdictions, however “lots of energy and cooling” is relative. A single rack would require a huge solar array, and either a huge passive radiator or an active radiator with heat pumps to radiate heat faster but requiring even more energy to power. If this data center is ever going to be occluded by the earth in orbit, then it either needs to power down or have enough battery backup to run without solar, which requires even more solar panels Presumably the entire system needs some kind of protection in the form of shields from solar flares/cosmic rays.
When it comes to networking here on earth we can hookup datacenters with relatively cheap fiber. Communicating with satellites posses more significant challenges. There’s nothing we can do about latency, which will be poor no matter what. You’d need many ground stations around the world with large radio dishes to communicate with satellites.
I guess such a system would technically be feasible given enough money, but the big problem is really bandwidth. It’s never going to be as scalable as an earth based data center.
It’s a fun thought experiment, but I think you’d be better off with an ocean based data center. Incidentally big tech companies have been building floating data centers for decades.
https://www.theguardian.com/technology/2013/oct/30/google-secret-floating-data-centers-california-maine
I don’t know if it is true, but in the past I’ve heard that one perk of such datacenters is to avoid paying taxes.
No one currently protects from solar flares/cosmic radiation. It’s just one of those things satellite companies are willing to roll the dice on. Well, I imagine national military spy satellites are rad-hardened, but nothing civilian.
Fair points mostly. I calculated (not easy) that a data center to handle all of the US requirements would take a solar / cooling combined area of 96 square miles. We can’t get that into orbit and it would be torn to shreds by micrometeorites.
However there is no need for batteries or more than one ground station. Geosynchronous satellites orbit at approx 26K miles, approx pie times the diameter of the early. So the earth’s shadow is avoidable. Only one ground station is needed, other than back-ups.
Agreed latency is OK for human interface, but not for any tightly coupled earth to space processing. Light (or microwaves) would 26/186 seconds to get to the earth.
lapx432,
Even with a geosynchronous or geostationary satellite, you’ll end up with the earth casting a shadow on the satellite sometimes. You could have a satellite in an orbit that doesn’t cross earth’s shadow for parts of the year, but the shadow vector itself is not stationary. It rotates around the earth as the earth rotates around the sun. At some point the shadow vector is going to cross the orbit of the satellite and it will end up in the shadow of the dark side of the earth. I need a satellite simulator to visualize it better.
IMHO you should provide more details as to how you arrived at your 96sq miles figure, but in any case thats about 11mile diameter circle. To be geosynchronous, it’s center of gravity has to be orbiting 22miles away from the earth’s surface. Due to how big it is, there would be a significant gravity gradient across the structure (at the most extreme, the near parts will be 16.5miles away while the far parts are 27.5miles away). This thing would be quite monstrous looking in our sky, occupying about 27degrees of our field of view by my rough estimate. It would also cast it’s own shadows on earth…though this may help with global warming, haha.
This is quite amusing to ponder 🙂
Alfman: ” At some point the shadow vector is going to cross the orbit of the satellite and it will end up in the shadow of the dark side of the earth. I need a satellite simulator to visualize it better. ”
I don’t think you need a simulator. The angle formed by the triangle between the line from the geo sync satellite to the center of the earth and the line from the geo synch satellite to the edge of the earthly is approx 9 degrees. Sin(-1) (4/26) = 8.8 degrees. The 23 degree tilt of the earth means that the satellite is outside the shadow of the earth when on the opposite side to the sun.
I think the only solution if you have to put data in space is micro data centers orbiting close to the earth in a ring with the data maintained as geosynchronous by having it transmit to the next data center at the speed of the rotation of the orbit. Your data center orbits but your data doesn’t. Those things would need batteries. I recommend counter rotating flywheels. They can be used for orientation as well as energy storage. This ring would also serve as a world wide network, except maybe at extreme northern latitudes. Now think of the distributed operating system required and we are back on topic with this post :-).
Also you would have redundancy as a satellite could be skipped if it went down.
The gravity shear is a great point. It would be a huge issue in keeping an array of that scale pointed to the sum. You would probably have to solve with a few thousand smaller stations linked power wise by microwaves. Highly complex given that can’t eve solve a three body problem.
I came up with the 96 square miles as follows. US data centres use 17 GWatts. Let’s assume 50% of this is for cooling so we need 8.5 GW for computing. Let’s assume free cooling via black radiators. The international space station generates approx 1 KW per 10MSq of solar panels. So the US space data centre needs 85 million square meters of solar panels or approx 32 square miles. Cooling needs to radiate the same energy as the energy retained by increases in entropy in the information in the data center is negligible. Let’s assume that radiant cooling is half as efficient as solar energy generation. Solar panels radiate about 70% of their energy out the back so this is not too far off. And this is also why we can’t use the back of the solar panels for cooling. So we need another 64 square miles of reflective front and black backed cooling arrays, i.e. a total of 96 square miles or a disk with a diameter of 10 miles.
lapx432,
I wrote my other post before reading this, but that post seems redundant now since it sounds like we agree that the cooling is not trivial. The heat radiators would be huge.
Anyways, a billionaire like bezos will probably put a mini data center in space at some point because he can. He might even sell some EC compute nodes on it, but I suspect it would be more of a novelty than anything practical. Everything would be more expensive and slower.
Coming back your original point, I’m not sure any of today’s tech entrepreneurs would be willing/able to protect users from government interference even if the datacenter were floating in space. The US and EU governments already disregard jurisdictions on earth, they may simply disregard autonomy in space as well.