In 2016, the ISS was still running on Intel 80386SX 20 MHz processors, which were already a quarter of a century old. GLONASS ground stations are equipped with the first version of Elbrus. In “small space”, the priorities are different: low cost, rapid iteration, and the use of CubeSats on Raspberry Pi and Linux containers. Let's take a look at why time-tested technologies are valued in space.
Why Everything Happens So Slowly
If you look at a space project through the eyes of engineers and managers, it resembles not a startup, but rather the construction of a nuclear power plant. Everything takes a long time, is multi-layered, and requires incredible precision.
The first stage is to define the goals and objectives of the mission. This is where the main idea, tasks, budget, and deadlines are formed. This stage includes researching technical feasibility and evaluating key technologies that need to be developed from scratch for the specific conditions of space. For example, radiation protection and long-distance communication systems.
Next comes preliminary design. During this stage, a preliminary appearance of the spacecraft is created, and technical specifications are developed for further work. And this is where the specifics begin. In a satellite, you can't just take and replace components like you would in a server in a data center. And you certainly can't just take and start using a new satellite if something goes wrong with the old one, as we do with our gadgets. Everything that the engineers have chosen will remain on board for the entire service life. They could have made their selection back in early 2010, but the launch is only happening now.
The next stage is detailed design. At this stage, the design, materials, and software are worked out, taking into account reliability requirements.
Next comes certification. This is one of the most rigorous and lengthy stages of the project life cycle. There is an international standard, DO-178C, which is a “code of conduct” for developers of space and aviation software. In accordance with this standard, no line of code can be formally approved until it has been absolutely proven that every element of the system — from the vaguest idea to the code in the microprocessor — has been traced, described, and tested.
Certification is a slow process: auditors check plans, tests, and documentation, right down to code coverage (in critical systems using the MC/DC method — Modified Condition/Decision Coverage). This process takes years and makes 7–10 years of preparation for launch the norm rather than the exception.
Then the testing phase begins. The device undergoes tests in conditions as close as possible to those in space—vacuum, low temperatures, vibration, and radiation. This is done to confirm its operability.
Next comes system integration and launch preparation. This also includes final certification and audits required for the device to be approved for launch. But it is this “bureaucratic marathon” that guarantees that the satellite will operate trouble-free for 15-20 years in conditions of radiation, temperature fluctuations, and without repair.
Then comes the launch itself. After that, the device is put into operation and begins working in space. During this period, communication is maintained, software updates are downloaded (with significant restrictions), and orbit parameters are monitored and adjusted. All this is done through ground stations, with multi-stage verification to ensure that an accidental bug does not disable a mission worth hundreds of millions of dollars.
The Telescope Incident
However, in exceptional cases, repairs and modifications are still possible directly in space. For example, in 1990, the Hubble telescope was launched into Earth's orbit. It was sent beyond our planet to avoid distortions caused by the Earth's atmosphere when collecting data. Ground-based observatories constantly face problems such as air turbulence and the absorption of ultraviolet and infrared radiation.
A few weeks after launch, it turned out that the telescope's main mirror had spherical aberration — a deviation in shape of 2.2 microns (less than the thickness of a human hair). This caused the image to become blurred.
The cause of the malfunction was incorrect equipment settings during mirror polishing. In 1993, one of the most difficult repair missions in the history of manned spaceflight took place. Astronauts traveled to Hubble and installed the COSTAR system, a set of lenses that compensates for aberration. At the same time, they replaced the main camera with a new one, already equipped with built-in correction. Thus, Hubble finally began to produce clear images.
After the success of the mission, NASA conducted four more expeditions.
- In 1997, they replaced the GHRS spectrograph with a more powerful STIS spectrograph with high sensitivity, and also added NICMOS, a camera for infrared observations cooled by liquid nitrogen.
- In 1999, they replaced all six gyroscopes and upgraded the main onboard computer.
- In 2002, the ACS camera was installed, and the solar panels were replaced with new ones.
- In 2009, the WFC3 camera was installed, and all gyroscopes and electronic units were replaced. The STIS spectrograph, which had failed in 2004, was also repaired.
And yet, this case is an exception. It is important to remember that if your satellite is not on par with Hubble in terms of price and scientific significance, no one will fly up to repair it. Incidentally, the telescope's service life has been extended from the planned 15 years to more than 35 years thanks to upgrades. Typically, satellites operate for up to 20 years.
What happens if a spacecraft is left in orbit for too long?
Among these old-timers is Transit 5B-5, launched by the United States on December 21, 1964. It is still working. More precisely, it transmits signals — mainly for scientific and testing purposes. This machine from the era of the first navigation systems runs on the radioactive isotope plutonium-238, which is why it has been orbiting for so long.
An interesting fact: Transit 5B-5 became the direct ancestor of GPS. It is an example of how the simplest electronics of the 1960s, with a reliable power supply and resistance to extreme conditions, can operate in orbit many times longer than most modern satellites.
At the end of its life cycle, the device is either sent to a “graveyard orbit” or sent to the bottom of the ocean in a controlled manner. According to estimates by the European Space Agency (ESA), there are more than 54,000 objects larger than 10 centimeters flying above our heads. There are already about 1.2 million small fragments measuring 1 to 10 centimeters. And the number of sub-centimeter fragments has long exceeded 130 million.
By the beginning of 2025, there were about 40,000 satellites in Earth's orbit, and only 11,000 of them are operational. If the density of space debris continues to grow, it could cause the Kessler effect, rendering near space unusable. Even a tiny grain of sand flying at a speed of several thousand or tens of thousands of kilometers per hour could irreparably damage the Hubble Space Telescope or the ISS life support system.
Scientists are seriously concerned about debris in orbit. It needs to be cleaned up, but where?
Currently, there is a secluded spot in the Pacific Ocean for this purpose — Point Nemo. It is located in the southern hemisphere, approximately 4,800 km from the coast of New Zealand and approximately 2,700 km from the nearest islands. This is the so-called “spaceship graveyard,” where debris from spacecraft that did not burn up in the atmosphere is sunk.
At Point Nemo, the ocean is about 4 km deep, with low nutrient content and virtually no life. When the ISS reaches the end of its operational life, it will be brought into orbit over the ocean and sunk there.
To be more precise in terms of numbers: the time from the start of preliminary design to launch usually takes 7-10 years (Sentinel-1A: approval - 2007, launch - 2014), followed by 15-20 years (or even more) of operation, which correlates well with the current programs of NASA and ESA.
This creates an extremely paradoxical situation: spacecraft use technology from the last century, but this strategy is what makes space reliable. As a result, satellites designed and certified to 1990s standards continue to provide communications, navigation, and scientific data for decades, while our terrestrial gadgets have been replaced several times over.
Radio and Thermal Resistance of Proven Standards
Processors and computing systems used in space, especially those on board satellites and interplanetary spacecraft, must have high radiation and thermal resistance. In conventional commercial electronics, these characteristics are considered excessive, but for space, they are vital.
Take the legendary BAE RAD750 microprocessor, based on IBM PowerPC 750 architecture and developed using 250 nm (or 150 nm) CMOS technology. It operates at frequencies from 110 to 200 MHz, provides up to 400 MIPS of computing power, and consumes about 5 W (or 10 W as part of a single-board system). It is one of the most popular and proven.
This is achieved through special crystal design, isolation, and data recoding, as well as careful component selection and testing for exposure to space radiation in conditions close to those of actual missions. The RAD750 is used in satellite and interplanetary probe control systems, such as the Curiosity and Perseverance rovers, as well as in telescopes.
In the USSR, one of the first Soviet onboard computers was the Argon-11S. It was the world's first space computer. It had triple hardware redundancy and automatically controlled the space flight according to the Zond program (a flight around the Moon with the return of the landing module to Earth).
A distinctive feature of these Soviet and Russian systems is that they were developed according to more conservative technological standards using less dense technological processes — for example, 0.18 μm. This increases their resistance to radiation and reduces the risk of failure. Although outdated by modern standards, these technologies have been proven over decades of operation in space. They can withstand conditions of intense radiation, extreme temperature fluctuations, and long periods of operation without the possibility of maintenance or repair.
RTOS and Languages
If everything is so complicated, is it really possible to send a computer/server running Windows or Linux into space? In theory, yes, but usually, such tasks require an RTOS — an operating system that guarantees the execution of critical functions without the slightest failure or delay. Among the most well-known and trusted RTOSs are the American VxWorks and RTEMS.
VxWorks, developed by Wind River, is a commercial RTOS with a high degree of reliability and numerous features. It supports multitasking with priority preemption and provides minimal response times. The OS is used by NASA, as well as in European and American satellites and scientific instruments. VxWorks has a modular architecture and is certified to aviation and space safety standards. The latest versions even integrate capabilities for working with AI and service containerization.
An open source alternative is RTEMS (Real-Time Executive for Multiprocessor Systems). It was originally developed to control US Army missile systems and was later adapted for multiprocessor architectures. The European Space Agency actively uses RTEMS because it is easily portable to different hardware platforms, including the radiation-resistant SPARC LEON family of processors, which are widely used in European missions.
RTEMS has a more flexible task scheduling system and allows for component modifications. The OS has undergone rigorous testing and has been awarded a reliability level of “B” according to the ESA classification, which indicates its suitability for critical space systems.
There is also Ada95, a programming language created in the US in 1980 for critical software in real-time systems. Like the RTEMS operating system, it was originally a military development that was adapted for scientific tasks. Ada95 is used in aviation and space due to its strict typing, support for parallelism, runtime array boundary checks, and exception handling.
Specialized software called “C” provides low-level control of response time and memory, so in safety projects, it is restricted by profiles such as MISRA C and supplemented by rigorous static analysis.
For certified real-time, there is the Ravenscar profile, which cuts tasks down to an analyzable subset. SPARK, a subset of Ada compatible with DO-178C/DO-333 requirements, is used for formal verification. In practice, these tools significantly reduce the amount of verification and operational risks in large projects, from fly-by-wire systems such as the Boeing 777 to ESA avionics.
CubeSat and COTS Revolution
There is currently a significant increase in interest in small satellites around the world. This is partly due to the revolution brought about by CubeSats. These satellites are slightly larger than a Rubik's cube, measuring 10×10×10 cm and weighing no more than 1.33 kg.
Thanks to their small size and modularity, CubeSats simplify and reduce the cost of creating and launching space equipment into orbit. This opens up new opportunities for businesses and researchers. For example, to launch commercial and scientific missions at a lower cost, including communications, remote sensing, IoT, and experiments in low Earth orbit.
CubeSats have become one of the drivers of the mass democratization of access to space and the development of the modern space industry. This is happening, among other things, thanks to the launch of serial devices using commercially available equipment (COTS). This approach allows for the rapid creation of modular and inexpensive satellites for scientific, commercial, and industrial tasks.