Re:The space between theory and practice in distributed systems
by api on Sunday August 10, 2014 @02:24PM
(#8160301)
There is a parallel divide in AI: "neats" vs "scruffies". Neats like systems that are provably perfect and formal, while scruffies do stuff like multiple superposed algorithms with fudge factors in between that are set by a genetic algorithm. Formalists run away screaming but it works "sometimes" where sometimes is the problem domain they are working within.
"But it's not mathematically stable!"
"It has been for the last month..."
My suspicion, outside of johnparkerg's polarization, is that distributed systems in practice are particularly messy from a formal standpoint, in contrast to non-distributed systems, while in practice you can live with the messiness, if you can reduce it a sufficient number of 9's, which seems to be anathema to theoretical approaches.
For example, the proof of the CAP theorem is irrelevant in practice, specifically because no system I'm aware of makes the strong consistency assumptions that it requires. On the other hand, CAP behavior is definitely a problem, once you reach a certain size.