[karmakaze]

I get the point of choosing key metrics that are indicative of the reliability or quality of service. Having to narrow it down to a headline makes it less clear. For a datacentre region, counting number of outages may very well be a great bottom-line indication, with more fine grained objectives being tracked.