Hacker News Reader

Comment by nkcmr | Hacker News Reader

Why do all aviation IT systems feel like they are so fragile?

It feels like not a single month can pass these days without some "Airline X has major outage and ruins tons of people's days!" story on the news.

What is the tech behind these frail systems?

View on HN

More modern systems are also down once in a while. With aviation it's just more visible because one system can affect a lot of people.

I'm not sure when ATC systems (which I would count NOTAM to be one of) have last failed, I always had the feeling they're very stable. It's usually airlines that struggle with their applications.

I never know how well it applies but headline outages always make me think of that “The Coming Software Apocalypse” Atlantic article from a few years back[0]

[0] https://www.theatlantic.com/technology/archive/2017/09/savin...

All hyper optimized systems become fragile until they collapse.

The NOTAM system is ridiculous, no surprise it collapsed ... you can lose your ticket for flying thru a MOA that went active AFTER you took off.

What could possibly go wrong with a "push" notification system for every tiny little detail of aerospace scheduling? No force is as destructive as the good idea fairy.

For those who also didn't know, MOA: https://tfmlearning.faa.gov/publications/atpubs/AIR/air2501....

I'm not sure your specific example is a particularly good one. I can legally fly through an MOA that's active, without needing ATC or military authorization.

They need Kafka

They use async communication - they need Pulsar!

MOAs aren’t the same as restricted areas. You can fly in them if you want.

The risk of air-to-air collision goes up. Activities are supposed to stop when non-participants unwisely enter. Doing so may be impossible on short notice. https://en.wikipedia.org/wiki/2015_Moncks_Corner_mid-air_col...

That's the curse of niche software. Especially one made to fit, it's easy for penny pincher to decide "it works fine now, it doesn't need ongoing improvements" and you end up with some ancient shit needing java 1.7 (or worse) to run with no plan to upgrade.

The software is tied to a dependency chain of real people (passengers, crew, etc), real equipment, and lots of constraints. So problems have a downstream effect that's more pronounced and visible than many other spaces.

I think it's just a question of media visibility. Aviation outage = queues in airports = easy article to write

When was the last time NOTAMs went down? It seems remarkably resilient actually.

Airlines view technology as a cost center.

This is the FAA's system, not the airlines'.

GP asked about both systems. It's worth noting that the FAA's systems are brittle for an entirely different reason than, say, Southwest Airlines'.

Reply on news.ycombinator.com