Prompt reaction is critical.
I blame Vanderferker, he didn't mention that I needed to run Girlfriend 2.0, Wife 4.0 and Mistress 3.1 on servers that were several parsecs apart.
Vanderferker lights up a Camel Blue... "Baby, that was hot. Alexa, activate fire suppression program one."
Which raises the question can you fry a terminal with a stunner? How hardened is an individual terminal? I think we're all accustomed to Hollywood tropes where the bridge is damaged, on fire, smoking, and sparks are flying everywhere. Makes for good dramatic sequences.
I would expect that they are space capable/hardened, so weapons that just disrupt human nervous systems would do little.
I have a post brewing on that, using a combination of the CT computer model damage system in starship combat and the LBB8 robot brains. Think 12 robot brains spread throughout the ship, each one of those computer hits are taking out one of the nodes but the other nodes stay up.
My take on it, the change is capable of being done in plank time, but propagates at C....
Fibre optic backup will tend to resolve a lot of radiation induced Single Event Upset scenarios, but even then you're probably wanting to look at a minimum triple redundancy architecture design in order to better resist this kind of "The Universe Hates Computers" sort of thing.
How do you account for all the deck plans showing a single computer adjacent to the bridge with the accompanying text also specifying as standard that main computers are located adjacent to the bridge?
The computer rooms would be the main console/admin location. All the other nodes are wedged into the walls/floors/ceilings supporting specific functions as well as backup synced processing.
As I understand it, modern avionics systems are redundant, with 3 separate systems and they "vote" on decisions. The Space Shuttle also had a voting architecture. Voting is when all 3 systems report a value, and the supervisor compares them and takes the majority view as the correct "answer". Who supervises the supervisor? No idea!
That sort of thing is what I expect my nodes are doing, voting and if enough don’t vote nothing happens (the modeled result when a CT computer phase check fails).As I understand it, modern avionics systems are redundant, with 3 separate systems and they "vote" on decisions. The Space Shuttle also had a voting architecture. Voting is when all 3 systems report a value, and the supervisor compares them and takes the majority view as the correct "answer". Who supervises the supervisor? No idea!
Back in the day, they were thinking of doing 3 clean room implementations on 3 separate CPUs, using 3 different computer languages, to minimize some commonality sneaking in among the systems. As I understand it, they tabled that, but do I think they did 3 clean room implementations to a shared spec, but used the same processor and language.
WRT the TU, the starships simply don't suffer this problem as weight is not a substantial problem with spacecraft, so shielding the computers appropriately will be less of a problem.
That said, this is an actual problem today. Story is that eBay would record some values more than once (like bid prices) in their database to ensure the rogue bit flips did not happen. I, myself, am confident that I have seen the aftermath of something like this once during my time developing software. Today, we have things like ECC memory to help automatically correct those issues, but, for example, my machine has 72G of RAM, but it's not ECC. The odds of my system encountering a rogue bit flip are fairly high (the impact of such, perhaps, not so much, but the possibility of the event, for sure).
Yup. Defense in depth.
STS used a quintuple redundant system, not a triple.
