The gap between the rhetoric of five-nines and actual network performance leads to the conclusion that five-nines may not be a realistic or even necessary goal.

In my experience, the biggest problem is the mismatched expectation: Marketing (getting their data from Engineering) proudly trumpets this performance, but defines it as an AVERAGE over the entire installed base. Each customer, however, assumes it means a guarantee for himself alone. You can't have it both ways. In fact, my employer has analyzed such data from that part of the installed base that reports back home, and in fact they claim 99.999% overall. (I wasn't privy to the definition and selection of outliers, but I'll bet there are some.) Not a network, just a bunch of boxen, but still. And remember: half the population[1] has an IQ below 100. :-) [1] Yes, I know the difference between mean and median, but that's not funny.