YoMomma ·
Nothing? 😬
SomeMichGuy · M
You mean systems such as the multiple independent backup systems on aircraft which are all supposed to be verified working before a flight?
Hireath · 36-40, M
@SomeMichGuy Yes
SomeMichGuy · M
@Hireath I know that Northworst would fly without a 2nd backup in operating order, at one time.
SomeMichGuy · M
@Hireath To be effective, they need to be actually independent.
The T3 NSFNet backbone had a logical diagram which looked nice, but the physical routing of some circuits ended up being in the same locations for certain segments (making them NOT independent).
As a result, TWICE a "backhoe in the FL panhandle" took out TWO links.
I'd not be surprised if some sort of failure were a common/most-common way of discovering non-independence.
Therefore, desigining it in has to be guided by contemplated failure modes; if a pie or pareto chart can be developed to show most likely failure modes, it ought to be possible to create better isolation of
• possible failure vectors
• potentially failed controls / components, and their location(s)
• activation/control pathways for alternate controls and/or alternate components
MULTIPLE redundant systems clearly make this a much more difficult problem.
But it suggests a common methodology based upon hyper-awareness of actual physical locations and routings, and stringent isolation.
The T3 NSFNet backbone had a logical diagram which looked nice, but the physical routing of some circuits ended up being in the same locations for certain segments (making them NOT independent).
As a result, TWICE a "backhoe in the FL panhandle" took out TWO links.
I'd not be surprised if some sort of failure were a common/most-common way of discovering non-independence.
Therefore, desigining it in has to be guided by contemplated failure modes; if a pie or pareto chart can be developed to show most likely failure modes, it ought to be possible to create better isolation of
• possible failure vectors
• potentially failed controls / components, and their location(s)
• activation/control pathways for alternate controls and/or alternate components
MULTIPLE redundant systems clearly make this a much more difficult problem.
But it suggests a common methodology based upon hyper-awareness of actual physical locations and routings, and stringent isolation.
dirge · M
enough to be annoying, not enough for it to be profitable





