As everyone in the software world knows, debugging, especially in production, is best described as frustrating, time-consuming, and often risky. Now, imagine debugging a device launched in 1977, currently 15 billion miles away (and counting) where every command sent takes around 22 hours to reach, and another 22 hours for the response to come back.
All in a day’s work—or two, to be honest—for the engineers at NASA’s JPL managing Voyager 1.
It’s truly mind-blowing when you consider the logistics involved and how forward-thinking the original mid-1970s engineers of the software/hardware had to be to make this device so accessible.
The software sits on a chip, naturally, but they built in what is effectively a back-door, allowing them to modify the program mid-flight. This isn’t just a simple update routine but the ability to change memory values. Talk about playing with live ammo.
Getting this wrong is unfathomable. You can’t just “reset” here, or bring the device back for a reboot. This thing is hurtling away from you at around 38,000 mph, running on 1970s technology. Incidentally, this wouldn’t have been cutting-edge at the time, as NASA does not risk unproven technology—it was likely cutting-edge in the late 1960s.
The “staging” environment for Voyager 1 must be something to behold, as every little step must be carefully tested and retested to minimize the risk of failure. However, even this staging could be out of step with the actual chipset, as it has not been subjected to 46 years of deep space radiation, and anything could have happened since the last command was sent.
Next time you’re cursing that your CI/CD is too slow or holding you back, spare a thought if your application was literally out of this world.
PS Take a moment and read about the computing power on board Voyager – it is humbling and impressive at the same time.
Alan Williamson 