Latest / Technology Now / How a supercomputer on the International Space Station is revolutionising science on the edge
Transcript
- Michael Bird (00:09): Hello, and welcome back to Technology Now, a weekly show from Hewlett Packard Enterprise, where we take what's happening in the world and explore how it's changing the way organizations are using technology. I'm your host, Michael Bird, as Aubrey Lovell is taking a break this week. Now, in the next couple of episodes, we're going to be looking at Spaceborne Computer-2, a supercomputer on the International Space Station that's revolutionizing computing at the edge and science in space. This week, we'll be looking at the science behind the computer, and next week, we'll be looking at the way it's changing the way research is conducted in orbit and beyond. So how is edge computing changing how scientists and astronauts benefit from space exploration? And what does it mean for you and me in our day-to-day lives? So if you're the kind of person who needs to know why what's going on in the world matters to your organization, then this podcast is for you. And if you haven't yet, do make sure you subscribe on your podcast app of choice so you don't miss out. Right. Let's get into it. (01:18): In 2017, HPE and NASA sent the first Spaceborne Computer to dock with the International Space Station. That mission to see how a supercomputer could cope with space also pioneered how AI can be used for future missions to the moon, Mars and maybe even beyond. The mission was followed by Spaceborne Computer-2, and then in January, 2024, an upgrade to parts of the system left Earth after being returned to the surface in 2023. This time around, the mission focused on allowing astronauts to carry out real-time data processing on board by using edge computing for the first time in space. But there's more. It's the first time a computer built from off-the-shelf parts has been used in space, which, if it works without being affected by radiation, could lead to much cheaper space missions in the future. (02:09): So what can Spaceborne Computer-2 teach us that wasn't possible with previous missions? And what difference is having a super computer on board making to the astronauts on the International Space Station? To fill us in is a man who assembled the virtual team for the mission and has developed the payload software for Spaceborne Computer-2, or SBC-2, principal investigator for Spaceborne Computer-2 at HPE, Mark Fernandez. So Mark, thank you for joining us. Just quickly, what is Spaceborne Computer-2? Can you give us a quick history of the program? Mark Fernandez (02:45): Well, Spaceborne Computer was a vision from NASA Ames who understood that when humanity gets to the moon and gets to Mars, we're going to have to take some compute resources with us. And so those visionaries asked us to take one of their compute nodes from one of their existing supercomputers and see, A, if we could fit it into a rocket, and B, could it survive the shake, rattle and roll of launch, and thirdly, could these non-IT trained professionals called astronauts install it and get it up and running. So that's what we've done. And our Spaceborne Computer-1 was a proof of concept to see if we could actually do that, and the focus was on the hardware. With Spaceborne Computer-2, our focus has expanded and probably our emphasis now is on proving out the value of workloads at the edge of the edge on the International Space Station. Michael Bird (03:50): Okay. So we've sent up Spaceborne Computer-1 and Spaceborne Computer-2, but [inaudible 00:03:58] saying there is now a upgrade to the Spaceborne Computer-2? Mark Fernandez (04:02): Yes, indeed. NASA allowed us to have a mid mission refresh, if you would. So we brought down half of Spaceborne Computer-2 and we did, quote, unquote, upgrades to it. The major upgrades were software, firmware, et cetera, also networking and security, but we're able to add quantity four 30 terabytes solid state disc. So I have 120 terabytes of storage at the edge of the edge on the space station. I can pretty much handle any data storage needs that typically come across. Michael Bird (04:42): So one of the key components of Spaceborne Computer is that you can use off-the-shelf parts. Essentially I can replicate what is on the International Space Station. Why haven't we just created some fancy bespoke system that would potentially be more refined for the requirements of being up on the International Space Station? Mark Fernandez (05:06): So you want to stay in step with the industry and HPE does that. So we want to take what HPE is mass producing with its quality control, et cetera, for use at the edge and put that on the International Space Station. We take one of them right off the factory floor. Just like you said, you could order one and take it home. Secondly, what are the requirements of every single scientist and every single vertical across the entire world that's going to be using equipment in space? We can't define that. So we trust that HPE has done that for, quote, unquote, general purpose edge computing, and that's what we're sending up there. So we've been able to address healthcare, life sciences, manufacturing, telecommunications, et cetera, all of these different verticals with these quite capable systems. Michael Bird (06:03): And so having off-the-shelf computer hardware in space and on rockets, presumably there are some challenges with the hardware. Does it cause the hardware to fail? Mark Fernandez (06:16): That was the purpose per se of Spaceborne Computer-1 to see if we could do it, and if so, how long would it last. The goal, of course, is that modern computer scientists and engineers want to use modern technology, and the currently available IT equipment that you can take with you is 10 or more years old. So can we take the latest and greatest commercial off-the-shelf to space and have it function? And to date, we're proving that you can. So the radiation is still a concern, but we have attempted to mitigate that with standard redundancy practices, hardware redundancy and software redundancy, et cetera. Michael Bird (07:03): So what is it in particular about launching something on a rocket or putting it in space? That means it's a particularly hostile environment for off-the-shelf hardware compared to just sat in a data center somewhere in a city? Mark Fernandez (07:18): Standard data center is climate controlled and it has conditioned electrical power coming into it and it has backup power generators, et cetera. Well, when we're on the space station, the electrical power is not that clean and not that reliable. We are very pleased with the water cooling. We were allowed to tap into what's called the medium temperature loop on the International Space Station, and we have a miniature rear water cool door as part of Spaceborne Computer. And the cooling has been outstanding. Networking is the third key factor. In data centers here on Earth, you've got multiple redundant high-speed networks back to your office or your building or your home office. Well, we all share a very tiny straw back to Earth. So we are exploring and proving out ways to encode, encrypt and compress your data up and down from space to Earth that really gives an advantage to those that want to do the edge computing and get their insights back to Earth as soon as possible. Michael Bird (08:28): Yeah. I'm guessing there's not a very, very long ethernet cable that goes from Earth to the International Space Station. Presumably, it's using some sort of microwave or radio transmitter? Mark Fernandez (08:38): Yes. They're using what's called the [inaudible 00:08:41] satellites, which have been up for decades. And so they're really, really reliable, but they're also incredibly slow relative to today's internet speeds. For example, at your house, you probably have 400 megabits or maybe a gigabit per second. Well, we've got two megabits a second coming down to Earth. Michael Bird (09:05): Brilliant. Thank you so much, Mark. And we'll be back with you in a moment. So don't go anywhere. Okay. Well, it's time for Today I Learned, the part of the show where we take a look at something happening in the world that we think you should know about. So this week, the news we have for you is that researchers from a university in the State of New York have created a way to speed up the development of 6G wireless. It's all down to a semiconductor chip that will enable smaller devices to operate at higher frequencies. The tricky bit in stepping up from 5G to 6G so far has been how the data is relayed. Whereas 5G runs below 6 gigahertz, 6G is expected to run at frequencies above 20 gigahertz, making it 100 times faster. So where the issue comes in is that data runs a higher risk of being lost the higher the frequency is. This is because the signal is unable to penetrate as deeply into the conductive materials on the chip. (10:01): So according to the research paper, snappily called Ultra-compact quasi-true time delay for boosting wireless channel capacity, the solution is to give the data more time to get from A to. B writing in the journal Nature, the research team led by Bal Govind working with co-author Thomas Tapen explained they have created a new chip that creates a time delay so multiple signals can come together at a single point in space without disintegrating. The resulting circuit almost doubles the data rate of existing wireless arrays without being any larger. And by getting all of this onto a smaller chip, it means 6G will be able to be used on our smartphones and IoT devices. But as to when that will be, well, watch this space. (10:46): Right. Well, it's about time to return to our guest, Mark Fernandez, to talk about the amazing science behind Spaceborne Computer-2. So we talked a bit earlier about how space is a pretty hostile environment for really anything. There's less atmosphere and so you're bombarded with more radiation, which can then have an impact on your computers. And I think there's this concept which would be great for you to explain to me called radiation hardening. Can you just explain what that is and why it's something that's important? Mark Fernandez (11:24): So yes, I can attempt to explain radiation hardening. I'm by no means a radiation expert. So experts will look at your electronics and come up with a way to protect them from the radiation. And this takes millions of dollars and tens of years to do. And you end up with something like a $200,000, two core processor. Michael Bird (11:49): Wow. Mark Fernandez (11:50): And that is the standard now that you would use in a satellite or a rocket, et cetera. But we scientists and engineers can't use that, correct? So we need something more modern. So we at HPE and the Spaceborne Computer team have created this concept called hardening with software. And in a nutshell, we have as much hardware redundancy as the commercially off-the-shelf systems will support and we have duplicate systems. So there's a buddy system. And then for each of those systems, we monitor every single parameter that for which there is a sensor and we keep track of whether or not those parameters are inbounds or out-of-bounds. And if one of them goes out of bounds, we ask our buddy, "Are you seeing the same thing?" And if he's not, then we defer to the buddy as probably the one with the correct answer. Michael Bird (12:47): I believe that certain kinds of radiation hitting computer components can cause something called bit flipping. Can you just quickly explain what that is? And why it can cause issues with computers? Mark Fernandez (12:55): Oh, absolutely. So everything in a computer is represented with bits, zeros and ones. And so you can take any string of bits, zeros and ones, and flipping them means if it's a one, it becomes a zero, if it's a zero, it becomes a one, and then it's something totally different and you'll get the wrong answer. So you want to avoid that. Fortunately, most modern components have what they call error checking on board. And so they will check for bit flips between their transmissions between memory and CPU or between the CPU and storage, et cetera. And if they notice something out of whack, they will do a retransmit and then they determine, "Okay. This is now good to go." So I think we're benefiting a lot from pure modern technology and it's error correcting that we have indicated on Earth that we need. Michael Bird (13:54): Wow. That's pretty fascinating. So how do you test the effectiveness and longevity of the system? Mark Fernandez (14:00): Okay. NASA and most people in the public know about the twins experiment. We sent up twin astronauts. We sent one to the International Space Station and one stayed on Earth, and we monitored their health for the entire mission, correct? And there are some significant findings from that twin study. Well, we follow that model. We have twins here on Earth and they're called Earthborne Computers as opposed to Spaceborne Computers. They're the identical computers, identical hardware, identical software, everything. And I run partners experiments on Earth at the same time I'm running them on space. And so I get to compare the answers on Earth and on space. And of course in space, I've got multiple copies. So I compare the copies in space before we actually release it to the partner and let them celebrate that they've done something successfully. So we're comparing the runtime, how long did it take to run, any errors they may have found, and the answers. And so I'm pretty confident when I get the right answer it's right if it matches the answer on Earth. Michael Bird (15:09): So what's next for the Spaceborne Computer program? Is there going to be a Spaceborne Computer-3 or a Spaceborne Computer-4? Mark Fernandez (15:16): Internally, we have settled upon staying with the edge line product line, and we're probably going to migrate from the Edgeline 4000, which we have now, to the more modern, more compact, more capable Edgeline 8000. Michael Bird (15:36): And so would that be called Spaceborne Computer-3? Is that the rough plan? Mark Fernandez (15:37): That's correct. Spaceborne 3-X and X could indicate where we're going. So -Lunar, -ISS, dash fill in the blank. Michael Bird (15:50): Wow. Gosh, that is very exciting. And is there a lunar mission planned, a manned lunar mission planned in the next few years? Mark Fernandez (15:59): Yes. Of course you probably have heard of Artemis. That's the big NASA plan to go there. And we're tracking that closely. There are also some lunar rovers and some, I'm going to call them, pre-placement missions to get ready for humans to arrive. And we're in discussions with the edge computing capabilities that those folks will need as they get there. Michael Bird (16:25): Gosh. So sort of a pre-supply mission with food and tools and other resources like that that go ahead of the manned mission? Mark Fernandez (16:34): Yes. Indeed. So we would like to stage a power infrastructure, oh, and we might need some ethernet, internet, 5G connectivity, oh, and we might need some edge computing processing because you can't get back to Earth as quickly as you can around here, and then you're going to need some computers to keep track of all of that infrastructure, so a command and control infrastructure, if you would, of which Spaceborne Computer could also participate in that. Michael Bird (17:07): Wow. Man, I mean, that is just so exciting. This sort of feels like the future. Does it feel like that for you? Do you feel like you're working on something that's going to have potentially a big impact on humanity? Mark Fernandez (17:17): Yes, indeed. And I often tell people that everything we do today here on Earth involves a computer. And we need that to be there on the next space stations, on the moon to support our exploration. We're no longer going to the moon just to say we went there and come back. We're going there to do something. And whatever it is you're going to do, you're going to need some edge computing capability Michael Bird (17:44): That is really exciting. Yeah. So you're sort of able to do things that people just didn't think were possible even five or 10 years ago and 20 years ago just would've been completely unfathomable because you just wouldn't have been able to process it. There just was no way there was enough processing power. And yet now that processing power is in space. Mark Fernandez (18:07): Correct. When people see the capabilities of edge computing on the space station or elsewhere, they got to shift their paradigms of how they've been doing things. We're no longer just doing things better, faster, cheaper. We're going to do it with a different paradigm, and it's going to be faster, better, and cheaper. Michael Bird (18:27): Mark, thank you so much. It has been brilliant to talk. And you can find more information on the topics discussed in today's episode in the show notes. Right. Well, we are getting towards the end of the show, which means, of course, it is time for this week in history, a look at the monumental events in the world of business and technology, which has changed our lives. (18:48): Now, the clue last week was, "Look sharp. In 1931, this invention created a buzz. Any ideas?" Well, it was of course, the first commercial electric razor released this week in 1931. Hooray. Now, Jacob Schick created the original electric razor in 1928, but it was a strange design where the motor was in a separate box and the razor head was on the end of a long wire. His 1932 design, on the other hand, was pretty much exactly what we see today and was a massive success. At the height of the Great Depression in his first year, he sold 3000 razors at $25 each, which is around $400 today. By 1937, he'd sold 1.5 million, creating an entire industry out of letting a little machine take the stress and nicks out of shaving. (19:45): Next week, the clue is, "It's 1963. Pew, pew, pew. Do you know what it is?" I don't know what it is. So I guess we'll be finding out next week together. That brings us to the end of Technology Now for this week. Thank you so much to our guest, principal investigator for Spaceborne Computer-2 at HPE, Mark Fernandez, and to you. Thank you so much for joining us. Technology Now is hosted by Aubrey Lovell and myself, Michael Bird. And this episode was produced by Sam [inaudible 00:20:14] and Al Booth, with production support from Harry Morton, Zoe Anderson, Alicia Kempson, Alison Paisley, Alyssa Mitri, Kamela Patel and Chloe Suewell. Our social editorial team is Rebecca Wissinger, Judy Ann Goldman, Katie Guarino, and our social media designers are Alejandra Garcia, Carlos Alberto Suarez, and [inaudible 00:20:33] Maldonado. Technology Now is a low street production for Hewlett Packard Enterprise. And we'll see you the same time next week. Cheers.