Architecture Consistency

Always keep your architecture design consistent, i.e. use the same solution for the same problem. If not, your architecture becomes unnecessarily complicated, which can lead to problems, sometimes even catastrophic ones. Here are two examples for such problems:

Apollo 13

You may have seen the movie with Tom Hanks about the “successful failure”, which was Apollo 13. One of the many problems they were facing was, that after they had moved from the Command Module to the Lunar Module to safe power for re-entry, the CO₂ levels were rising and there was the risk of dying because of CO₂ poisoning. The reason was that the lunar module was only developed for two astronauts to stay in there for 36 hours at the most. But there were three astronauts in there now and they had to stay for 96 hours. So they had to replace the air filters in the Lunar Module to get the CO₂ out again. However, the filters in the Lunar Module were round, while the spare ones they had for the Command Module were square. Nobody had even thought about an incident like that and therefore hadn’t cared about consistency in the design of those filters. Eventually NASA engineers came up with a fix for the problem and Apollo 13 made it back safely.

Here is the problem explained in a documentary about Apollo 13 and the corresponding scene from the Apollo 13 movie.

Mars Climate Orbiter

In 1998 NASA launched the Mars Climate Orbiter mission with the goal to study climate and atmosphere of Mars and act as communication node with the Mars Polar Lander. However, on September 23, 1999 communication with the orbiter was lost. The report of the Mars Climate Orbiter Mishap Investigation Board listed the reasons for the loss of the orbiter:

On September 8, 1999, the “Trajectory Correction Maneuver-4” for the Mars Climate Orbiter mission was computed and it was executed on September 15, 1999. The idea was to bring the spacecraft in a position that was perfect for another maneuver, which was supposed to bring the spacecraft into Mars orbit. That position was an altitude of 226 kilometers. However, shortly after the maneuver was executed, the navigation team recognized that the altitude was much lower at only 150 to 170 kilometers. And twenty four hours before the orbital insertion maneuver was supposed to start, they calculated the altitude at only 110 kilometers. Eventually they determined the altitude to have been only 57 kilometers, which means the orbiter must have been destroyed when entering the Mars atmosphere .

The cause was identified as an error made by the engineers. The flight system software developed for the Mars Climate Orbiter used the metric unit newtons (N) in thrust instructions. The software used on the ground to calculate the instructions used the Imperial measure pound-force (lbf). That caused a discrepancy between the calculated position and the measured position, which in the end led to the incorrect altitude.

This error is now known as “metric mixup” and NASA made sure to avoid it in all their missions since the incident.

Again, nobody made sure that the designs are consistent, i.e. in this case used the same unit of measure.

By the way, the Mars Polar Lander was also lost later. Here the Failure Review Board made the conclusion that the software had not accounted for vibrations of the legs of the Lander during entry into the atmosphere. So the software thought that these vibrations indicate a touchdown and turned the engines off even so the Lander was still about 40 meters above ground and the legs were still stowed.