Thursday, January 12, 2012

[iijbfuzf] Limits of fault tolerance

After attempting to account for all the possible failure modes you can think of, what is the probability of a failure that you didn't account for?  I believe this number can be estimated well from past examples of unrelated projects and measures of complexity.

Included in this "unknown" failure mode are the probability that your accounting of the known failure modes was wrong, e.g., a math error.  Also included is the human factor, which can induce infinitely creative new modes of failure.

Because of these unknown failures, there seems to be a limit to how safe something can be made, especially for one-off unique kinds of things.

Inspired by the biosafety level 4 lab in Boston.

No comments :