The International Space Station (ISS) is mankind’s most rarefied laboratory, a place where experiments can be performed under extended periods of true microgravity. During its more than two decades in operation, the ISS has hosted more than 2,000 experiments by scientists from around the globe, on topics ranging from cancer research and biotech to explorations of chemistry and physics. However, scientists on orbit have had to work without a key tool used in most modern laboratories: a sophisticated data center for processing their results.
The devices in the AWS Snow Family (which consists of Snowball and Snowcone) are designed to withstand remote and inhospitable environments, and it’s hard to imagine a more remote and inhospitable environment than space. After seven months of rigorous NASA testing, the AWS Snowcone required almost no modifications to go aboard, apart from being covered in Kapton tape (a just-in-case measure to ensure its surface materials didn’t interact with the possibly oxygen-rich atmosphere aboard the ISS).
—David Zuniga, senior director for In-Space Solutions, Axiom Space
“Being able to compute locally in space creates a significant reduction in data loads.”
Reducing latency is one of the most significant reasons researchers are eager to extend the edge to space. Data-transmission speeds to and from orbit are significantly slower than transmissions on Earth. Average ping times in the U.S. are around 100 milliseconds (ms), but latency to low-Earth orbit is around 375 ms one way (750ms round trip), with a much narrower data channel. From the moon, latency jumps to more than a second, while data takes up to 17 minutes to travel to or from Mars. Edge computing in space solves the same problem it does on Earth: It puts computing power close to the user, creating huge gains in speed and efficiency.
AWS takes a giant leap toward bringing cloud infrastructure to low-earth orbit for the first time
Space: the cloud’s
final frontier
10-31-22 | aws
BY FASTCO WORKS
+
REDUCING LONG-DISTANCE LATENCY
As more practical applications of spaceflight emerge in the coming years, edge computing will provide crucial technology infrastructure for a wide variety of operations. Several different companies (including Axiom Space) are currently building independent space stations that are projected to become operational in the next 7 to 10 years, forming the backbone of what is being called the Low Earth Orbit Economy. While these stations will serve partially as laboratories like the ISS, they will also be transitioning into practical applications including manufacturing, as certain materials, including stem cells and artificial retinas, collapse under their own weight unless made in microgravity. Space manufacturing will also be used to build items in low-Earth orbit that are impractical to build on Earth and send up on a rocket, including new space stations and massive space telescopes.
As space manufacturing progresses and systems become more sophisticated, they will need everything modern Earth factories do: Private 5G networks; networks of IoT sensors; and fast, efficient hybrid edge computing and storage. It’s likely that AWS hybrid cloud and edge services will also serve as a digital lifeline for space tourists as that industry becomes more common. “People are going to bring their digital devices to space,” Zuniga says. “When you’re up there you’re going to want to access and use data just as you do on the ground.”
This initial slice of cloud infrastructure is only the beginning. While it was convenient that AWS engineers made the AWS Snowcone rugged enough to withstand spaceflight, the next set of hybrid cloud and edge computing devices that are taken to space will likely be purpose-built for off-planet use.
SPACE MANUFACTURING: MICROGRAVITY MATTERS
When Axiom Mission-1 (Ax-1) lifted off from Florida’s Kennedy Space Center in April of 2022, it made history not only for being the first all-private expedition to the ISS, but also because of the payload its scientists brought with them: an AWS Snowcone—a 4.5-pound data migration and computing device designed to work in rugged, remote, and mobile locations without reliable internet connections. Which is to say, it’s a piece of the AWS cloud that can be taken anywhere. It represents an important step for humankind deploying and using cloud infrastructure in outer space.
“Being able to compute locally in space creates a significant reduction in data loads,” says David Zuniga, senior director for In-Space Solutions at Axiom Space, a leading space infrastructure developer. “Instead of sending every image needed for object recognition down to earth to be processed, you can run a machine-learning model close to your data source and get answers immediately.”
The Axiom-AWS team proved they could use the AWS Snowball to run a machine-learning object recognition model on an Amazon EC2 instance—a virtual computing environment—locally on the station. When Ax-1 departed for Earth, the AWS Snowcone stayed behind, and has since been reconfigured to run other models for different experiments. “It’s a very secure box from a data perspective, and it supports a select set of AWS compute and storage services,” says Daryl Schuck, business development manager for Aerospace and Satellite at AWS. “But the bigger opportunity is the ability to adapt the device to do what’s next. That’s the kind of flexibility and reconfigurability you’re going to need on a space station.”
AWS takes a giant leap toward bringing cloud infrastructure to low-earth orbit for the first time
Space: the cloud's final frontier
10-31-22 | aws
BY FASTCO WORKS
The International Space Station (ISS) is mankind’s most rarefied laboratory, a place where experiments can be performed under extended periods of true microgravity. During its more than two decades in operation, the ISS has hosted more than 2,000 experiments by scientists from around the globe, on topics ranging from cancer research and biotech to explorations of chemistry and physics. However, scientists on orbit have had to work without a key tool used in most modern laboratories: a sophisticated data center for processing their results.
When Axiom Mission-1 (Ax-1) lifted off from Florida’s Kennedy Space Center in April of 2022, it made history not only for being the first all-private expedition to the ISS, but also because of the payload its scientists brought with them: an AWS Snowcone—a 4.5-pound data migration and computing device designed to work in rugged, remote, and mobile locations without reliable internet connections. Which is to say, it’s a piece of the AWS cloud that can be taken anywhere. It represents an important step for humankind deploying and using cloud infrastructure in outer space.
The devices in the AWS Snow Family (which consists of Snowball and Snowcone) are designed to withstand remote and inhospitable environments, and it’s hard to imagine a more remote and inhospitable environment than space. After seven months of rigorous NASA testing, the AWS Snowcone required almost no modifications to go aboard, apart from being covered in Kapton tape (a just-in-case measure to ensure its surface materials didn’t interact with the possibly oxygen-rich atmosphere aboard the ISS).
—WAYNE CARTER, VICE PRESIDENT OF ENGINEERING, COUCHBASE
LATENCY IS THE ENEMY. IT CAN BE A MAJOR DRAG ON YOUR BOTTOM LINE. GETTING DOWN TO SINGLE-DIGIT–MILLISECOND LATENCY IS THE GOAL.
Reducing latency is one of the most significant reasons researchers are eager to extend the edge to space. Data-transmission speeds to and from orbit are significantly slower than transmissions on Earth. Average ping times in the U.S. are around 100 milliseconds (ms), but latency to low-Earth orbit is around 375 ms one way (750ms round trip), with a much narrower data channel. From the moon, latency jumps to more than a second, while data takes up to 17 minutes to travel to or from Mars. Edge computing in space solves the same problem it does on Earth: It puts computing power close to the user, creating huge gains in speed and efficiency.
“Being able to compute locally in space creates a significant reduction in data loads,” says David Zuniga, senior director for In-Space Solutions at Axiom Space, a leading space infrastructure developer. “Instead of sending every image needed for object recognition down to earth to be processed, you can run a machine-learning model close to your data source and get answers immediately.”
The Axiom-AWS team proved they could use the AWS Snowball to run a machine-learning object recognition model on an Amazon EC2 instance—a virtual computing environment—locally on the station. When Ax-1 departed for Earth, the AWS Snowcone stayed behind, and has since been reconfigured to run other models for different experiments. “It’s a very secure box from a data perspective, and it supports a select set of AWS compute and storage services,” says Daryl Schuck, business development manager for Aerospace and Satellite at AWS. “But the bigger opportunity is the ability to adapt the device to do what’s next. That’s the kind of flexibility and reconfigurability you’re going to need on a space station.”
BY FASTCO WORKS
REDUCING LONG-DISTANCE LATENCY
As more practical applications of spaceflight emerge in the coming years, edge computing will provide crucial technology infrastructure for a wide variety of operations. Several different companies (including Axiom Space) are currently building independent space stations that are projected to become operational in the next 7 to 10 years, forming the backbone of what is being called the Low Earth Orbit Economy. While these stations will serve partially as laboratories like the ISS, they will also be transitioning into practical applications including manufacturing, as certain materials, including stem cells and artificial retinas, collapse under their own weight unless made in microgravity. Space manufacturing will also be used to build items in low-Earth orbit that are impractical to build on Earth and send up on a rocket, including new space stations and massive space telescopes.
As space manufacturing progresses and systems become more sophisticated, they will need everything modern Earth factories do: Private 5G networks; networks of IoT sensors; and fast, efficient hybrid edge computing and storage. It’s likely that AWS hybrid cloud and edge services will also serve as a digital lifeline for space tourists as that industry becomes more common. “People are going to bring their digital devices to space,” Zuniga says. “When you’re up there you’re going to want to access and use data just as you do on the ground.”
This initial slice of cloud infrastructure is only the beginning. While it was convenient that AWS engineers made the AWS Snowcone rugged enough to withstand spaceflight, the next set of hybrid cloud and edge computing devices that are taken to space will likely be purpose-built for off-planet use.
+
