Antenna dishes at NASA's Deep Space Network complex in Goldstone, California, photographed on Feb. 11, 2020.
The Voyager 2 space probe has had an extraordinary life. Launched in 1977, the machine has continued operating more than 40 years, traveling through millions of miles of interplanetary space, past Jupiter, Saturn, Uranus, and Neptune, with no possibility of hands-on repair. And it’s still going.
Voyager 2’s instruments and cameras yielded scientific breakthroughs at each planet. For example, it found evidence of a boiling ocean about 500 miles below Uranus’s top cloud surface.
In 2018, the spacecraft was poised at the edge of interstellar space—only the second ever to travel so far from home, giving NASA an opportunity to take unprecedented measurements of the Sun’s magnetic field.
Capturing precious data like that involves precise planning and highly reliable technology. Data from Voyager 2 is captured on Earth by the Deep Space Network, consisting of three antenna sites positioned 120 degrees apart, around the world, so that NASA can have a line of sight with almost any spacecraft in the sky as the Earth rotates. However, due to its southerly trajectory, Voyager 2 can communicate only with the DSN location near Canberra, Australia. In 2020, for example, the largest of those antennas was down for maintenance, meaning mission operators could receive data from Voyager 2 but could not send it commands.
So NASA engineers repurposed antennas at the Parkes Observatory, about 180 miles from Canberra, and routed data from Parkes to a DSN facility in Goldstone, California. Without a hitch, NASA’s network of Oracle Private Cloud Appliances (PCAs) continued to receive and process the data smoothly, and Voyager 2 continued to receive commands.
It was just one success in a long relationship between NASA and Oracle, starting when NASA standardized on Sun Microsystems servers in the 1980s, on Oracle Database in 2006, continuing through Oracle’s acquisition of Sun in 2010, and through to today. Oracle PCAs provide the underlying computing power, networking, and data storage for the DSN, a network of satellite dishes near Canberra; near Madrid, Spain; and Goldstone, which is managed by NASA’s Jet Propulsion Laboratory (JPL) in Southern California.
The DSN connects the Earth with 40 spacecraft outside of Earth’s orbit, including the $10 billion James Webb Space Telescope, two Voyager probes in interstellar space, and Mars rovers, orbiters, and probes. Previously, the DSN connected with the Cassini probe, which was intentionally plunged into Saturn as its fuel ran out, giving unprecedented information about that planet’s atmosphere. DSN also connects with New Horizon, which flew past Pluto and continues sending data from the distant Kuiper Belt, some 4.6 billion miles from Earth. DSN both collects data sent from the craft and sends messages guiding the spacecraft as needed. In the future, DSN will communicate with planned explorations of Venus, the Sun, and astronauts returning to the Moon.
NASA also uses DSN to track probes sent by other countries, in agreements with Europe, Japan, India, and the United Arab Emirates.
These 40 spacecraft are humanity’s eyes and ears for the solar system and beyond. (Watch a NASA video with more background on DSN.)
In addition to projects deep in the solar system, the Oracle PCAs are also used for Earth science. One PCA runs hydrology apps, using data from aerial and ground sensors to map the water in the western United States, including in aquifers, reservoirs, rivers, and lakes, as well as in snowpack. The data is used to compile a map to understand water availability and drought risks in the region.
That same Oracle PCA also powers NASA’s Solar System Treks website, which collects data from space probes and combines it into Geographic Information System (GIS) infrastructure to produce surface map displays for bodies of the solar system including the Moon, Mars, Mercury, the asteroids Vesta and Ceres, Saturn’s moon Titan, and Jupiter’s moon Europa.
|The Deep Space Network connects the Earth with 40 spacecraft outside of Earth’s orbit, including the $10 billion James Webb Space Telescope, two Voyager probes in interstellar space, and Mars rovers, orbiters, and probes.
Solar System Treks will become more important as NASA plans to put an autonomous rover on the Moon to drive about a thousand kilometers across the surface. For that project NASA will need excellent maps of the Moon’s surface, showing slope, lighting, heat, and more.
Additionally, NASA researchers are using Oracle PCA to give people easier access to what they’ve learned from all this space exploration. It will run an instance of Dataverse open source software from Harvard that’s modified for natural language processing (NLP) on scientific papers published by JPL scientists. Dataverse can be used to summarize papers, tally the number of citations, and answer questions such as, “What is the chemical composition of the atmosphere of Titan?” (It’s about 95% nitrogen and 5% methane, by the way.)
DSN began using PCAs in 2015, replacing about 300 Sun Microsystems servers. Instead of all that physical hardware, NASA now runs 300 virtual machines on seven appliances in five locations—Madrid, Canberra, Goldstone, JPL, and a test facility in Monrovia, California. The PCAs run Oracle Linux and Solaris 11 for x86. Each PCA is connected to an external Oracle ZFS Appliance for storage, with a cluster of SPARC servers, and the entire infrastructure is managed by Oracle Enterprise Manager.
Each PCA has up to 24 compute nodes, typically using 768 GB RAM per node. The bigger PCAs run more than 500 virtual machines. DSN uses Oracle Database, which it has been running since 2006.
DSN expects to complete the seven-year transition from Sun servers to PCA in June 2022. Operational consistency was the main reason for the years-long transition. By using the same hardware and software configuration on every PCA, NASA expects to improve security and reduce operational costs, making it easier to manage patches and other software. Previously, IT staff at each location managed servers locally, resulting in different configurations for servers at different locations, which made operations more difficult.
The Mars Rover Perseverance being assembled in the “clean room” at the Jet Propulsion Laboratory (JPL) in California in January, 2020. Perseverance is now operating on the surface of Mars, communicating with Earth through the NASA Deep Space Network (DSN).
NASA runs its own, customized software and needs to be careful with security patches and operating system updates to avoid conflicts that could lead to outages for the entire ground system. JPL needs to be careful about updating methodically and consistently, and testing before it deploys.
NASA starts with Solaris 11.4 and Linux 7.9, then removes unnecessary packages. IT runs the software through a set of scans to view and remove vulnerabilities, which gives JPL clean “gold images” of Solaris and Linux to deploy automatically for the 300 virtual machines on seven servers. NASA wrote a command called “provision” to instantiate a VM, automating the workflow and locking down the template, with frequent audits to be sure the software remains consistent.
Vulnerabilities can be fixed once, quickly, and applied everywhere. Upgrades are accomplished with virtually zero downtime.
Oracle’s global support is a great benefit for NASA. If a part fails in Madrid, Oracle sends the part directly to that location, and a local Oracle tech installs the replacement. Prior to deploying PCAs, remote administrators were on their own.
Consolidating on standard versions of PCA allows JPL to consolidate maintenance and provides the ability for sites to support each other remotely, as a single team. Site admins can log in and provide diagnostics and support remotely. NASA uses two-factor authentication, while PCAs are on private VLANs to help ensure security. The simplicity saves operating costs. At NASA, like any organization, the cost of hardware is a small component of the lifecycle cost of running the system. PCAs are preconfigured for storage and networking, so they don’t need to be configured manually.
Also, the PCA deployment cuts red tape when rolling out new servers. Deploying a new hardware server used to require two or three months of paperwork, then the server needed to be configured with IP addresses and security scans. Now, all of that is done in minutes using virtual machines instead of physical hardware.
Reliability will be even more essential for future missions. Artemis 1 is a planned uncrewed test flight of a spacecraft that will go to the Moon and back, sometime after August 2022. Two years after Artemis 1, Artemis 2 is planned to take a crew of astronauts to the Moon and back. Any outage in communications with astronauts is simply unacceptable.
NASA has seen disasters take data centers offline—a fire in Canberra, snowstorms in Madrid, and power outages in Goldstone—so NASA needs its infrastructure to be robust enough to continue working through those sorts of problems.
The opportunities—and challenges—are enormous. Until recently, humanity has been confined to a single speck in the vastness of space. For the first time, we’re sending flimsy spacecraft off that speck. Deep Space Network is essential to allow those explorers to tell the people at home what they found.
Photography: Courtesy of NASA/JPL-Caltech