Is it time to reconsider placing networked servers in space?
Keith Lofstrom, an electrical engineer famous for designing the Launch Loop to move humans and materials into orbit, has also developed Server Sky, a series of “thinsats” that, when clustered together, would form data centers in space.
While admitting that the concept is speculative, Lofstrom admitted by email that he’s eager to connect with other hardware designers and computational software designers. Lofstrom has worked on and off on Server Sky for several years, and the project has been publicly discussed in scientific journals.
Lofstrom’s thinsats are glass sheets covered with very thin indium phosphide solar cells and combined with thinned integrated circuit chips, radios with antennas, and three disks of electrochromic material that act as electrically switched mirrors/windows. In other words, they’re powered by solar cells, propelled and steered by light pressure, networked and located by microwaves, and cooled by deep space. Lofstrom’s current thinsats are 3 grams in mass and measure 0.05 mm thick, according to the design—thinner than a sheet of paper.
The idea would be to stack the thinsats together and launch them into orbit. Lofstrom notes that a communication satellite such as the HotBird 9 weighs 4880 kilograms (10,800 pounds) and comes with a 14kW array, and is launched by an Ariane 5 launch vehicle, which can put 10,500 kg of satellite and apogee kick motor into a geosynchronous transfer orbit. The planned m288 thinsat orbit is lower and a larger payload is possible, Lofstrom wrote: 4,200 kilograms is equivalent to 1,400,000 thinsats and 5.6 megawatts of electricity. Thus, a thinsat array can produce more than 400 times the power (and communication capacity) of typical comsats—the clustered thinsats eventually pay for themselves with the power they save.
“We are much further along than the people who talk about space-based solar power—their problem is ever so much harder, they’ve been talking about it for four decades, and the capability to deploy (giant cheap boosters and space construction) is gone,” Lofstrom wrote in an email. “I know where to buy the pieces and get stuff made—that will require money, but not much more than most chip developments. But I need to get some of the path dependent stuff settled first, so we don’t end up designing IPV4 when we should start with IPV6. I also need to get this into the public domain—a thicket of patents will slow it down and distort the results.”
However, Server Sky will have to solve two big problems: maneuvering and communications.
Each thinsat will contain three “electrochromic thrusters” that can be rendered alternately opaque or transparent by applying a charge. Normally, the thinsat is pointed straight at the sun, and the thrusters are electrically stimulated into transparency. If one or two of the disks are unstimulated, they turn reflective, and produce about half a nanonewton of thrust, Lofstrom’s design documents indicate. (Lofstrom describes how the thinsats need to be oriented on a detailed navigation page.) The tiny amount of thrust each disc produces also means that maneuvers will need to take place over a period of hours.
Communicating is perhaps even more important. Lofstrom believes that thinsats will have the ability to communicate with one another, in order to better orient. It’s not quite clear how the individual signals will be synchronized, but the downlink signal will be combined from each thinsat and sent via frequencies that can penetrate the Earth’s cloud cover.
The bottom line is that traditional data centers are consuming more and more power, and that alternatives must be considered. Pie in the sky? Or an intriguing possibility? It’s certainly science fiction for now.