Hewlett Packard Enterprise’s space station computer is in demand

by
Q&A with Mark Fernandez, principal investigator for HPE's Spaceborne Computer-2

Since traveling in February 2020 to the International Space Station, Spaceborne Computer-2 has completed 20 experiments focused on health care, communications, Earth observation and life sciences. Still, the queue for access to the off-the-shelf commercial computer linked to Microsoft’s Azure cloud keeps growing.

Mark Fernandez, principal investigator for Spaceborne Computer-2. Credit: HPE

Mark Fernandez, principal investigator for Spaceborne Computer-2, sees a promising future for space-based computing. He expects increasingly capable computers to be installed on satellites and housed in orbiting data centers in the coming years. Edge processors will crunch data on the moon, and NASA’s lunar Gateway will host advanced computing resources, Fernandez told SpaceNews.

Fernandez, who holds a doctorate in scientific computing from the University of Southern Mississippi, served as software payload developer for HPE’s original Spaceborne Computer, a supercomputer that reached ISS in August 2017 and returned to Earth a year and a half later in a SpaceX Dragon cargo capsule.

What do people mean when they talk about supercomputers in space?

Small clusters at the edge are positioned as supercomputers because they are more than just a tiny edge device. We called Spaceborne-1 a supercomputer because we did one teraflop of computation in space. That’s orders of magnitude more than anyone had ever done before.

What are you learning from Spaceborne Computer-2?

What is surprising to me is the diversity of the experiments. We have 39 experiments in the queue, and the number of experiments is growing.

We’re analyzing astronaut DNA. That one, in particular, is pleasing to me because the scientists had been waiting weeks or months to get this big DNA sequence down to Earth to analyze it. You can compare this big dataset to the big human genome, but you’re only interested in the mutations.

Well, it took us about 13 minutes of processing and then about two seconds to download it. Suddenly, the scientists said instead of monitoring the health of one astronaut every month, they could monitor the whole crew daily and get a better idea of when space travel is adversely affecting them.

We’re looking at how satellites communicate with each other. Different types of encryption, different types of protocols, different types of compression.

What gives you the most security and uses the least amount of energy?

A lot of experiments have to do with weather and disaster preparation. High-resolution imagery of storms and tornadoes are large data files. Basically, the first responders just want to know where the forest fire is. What is the track of the tornado? You can tell them that in just a few words.

Instead of pictures?

A picture takes forever to get down. We can process that. I want to know where it’s flooded and not flooded. I want to know if the interstate is passable or not.

Are you sending only the most valuable information to the ground?

That’s the first layer of the onion that we’re exploring. It’s an intelligent edge. We don’t want to push all the computation to the edge. We don’t want to push all the computation to the cloud. If I have a multi-step workflow, I can do two or three steps at the edge. But I’m far better off bursting that smaller, mid-workflow results to the cloud.

For example?

It goes back to astronaut DNA. Mutations are updated all the time in databases at the National Institute of Health and the National Cancer Institute. We have the cloud search those databases.

What’s the best approach for various types of data?

We’ve got some serious propeller-head scientists running things only on the cloud or only in space on Spaceborne Computer. They differentiate it. They run it only on the CPU, only on the GPU. They are coming up with guidelines.

People also talk about edge processing for satellite operations.

The analog is autonomous driving. Just as all the cars will be talking to each other, all these satellites will be talking to each other. One of them is going to raise their hand and say, “I’ve got good connectivity down to Earth. I’ll deliver that message.” Then, they all agree.

Hewlett Packard Enterprise’s Spaceborne Computer-2, sent to the International Space Station in February 2021, is linked to Microsoft’s Azure cloud through NASA and HPE ground stations. Credit: NASA

HPE established an alliance in 2019 with OrbitsEdge, a Florida startup with a satellite bus for sensitive electronics. Are you working to install HPE computers on OrbitsEdge satellites?

Yes, indeed. OrbitsEdge is putting up a satellite with multiple distinct computers from HPE. To you, it looks like your computer on your satellite. But they’re actually hosting multiple computers from multiple people completely firewalled off from each other because they are on physically separate devices. They can run whatever protocols they want and whatever communications they like.

How do you envision computing resources in cislunar space?

When we get to the moon, the data center and the high-performance computing will be orbiting the moon, and the outposts will be the edge.

What are the challenges ahead for space-based computing?

They’re all related to space exploration. Power, cooling and networking are not stable. Networking is the most unstable. There are multiple times a day [on ISS] when we don’t have connectivity. If this was your cellphone, you would go get a new provider. But the space station doesn’t have an option.

Where do you imagine this going in Earth orbit, on the moon and Mars?

If OrbitsEdge gets its proof-of-concept going and can have a multi-tenant satellite, the next logical step is a multi-tenant data center built out of larger satellites. OrbitsEdge focuses on power, cooling and networking. They’re leaving that compute to us.

On the moon, you would have low energy communication up to the Gateway. The Gateway will have the power, cooling and storage. A similar architecture is being considered for the Mars outpost.

Do space applications continually demand more computing resources like terrestrial applications?

Yes, you want it faster, you want better networking, and you want more power. No one has complained that they have plenty of Spaceborne Computer right now. They ask, “When can I get back on it?”

This article originally appeared in the January 2022 issue of SpaceNews magazine.