Good to see consideration of out of band management there: interesting use for a fax line. The emphasis on lab verification of changes is useful too. That being said, this sounds more of a wish list by someone who hasn't had to deal with either a) hard business realities or b) very complex network failures. I would hesitate to describe myself as any sort of network "guru", but I do build out and support data centres, offices, and WAN links of various sorts, including metro rings, and the biggest causes of long outages are usually poor design (usually too complex), poor software (Brocade gets a special mention here!), and carriers (no comment!): in that order. Doubling and tripling up redundancy can improve failure rates (although I should mention that most vendors won't load balance links properly with IGP's or port channels unless they can be divided by 2), but that ain't necessarily so. My worst outage involved a stray OSPF default screwing up a multi-homed site without out of band management: a single homed site would have had higher reliability over the same calendar year.