I posted on Wednesday regarding some long term network issues that I thought were raised by the NYC attacks. An additional one has occurred to me from some subsequent events. For organizations which are dependent on the net for day to day operations (despite the dot-com collapse, the number of these is steadily increasing) the acceptability of single homing is likely to have decreased significantly. I addressed this a bit with the significance of true separation in physical links that I brought up in the first mail. Beyond that, though, we have seen that whole facilities can be outright destroyed or fail due to cascading infrastructure and backups failures as in 25 Broadway (are your generators n+1 redundant...and do you stock spare water pumps for them?). This really means that except for truly extraordinary ISPs, no end customer can really be sure that an ISP is going to be able to maintain services through a major incident (or even a localized serious incident). Expect to see in the coming months many more end customers desiring to get truly dually homed, with redundant ISPs, redundant local loops in different directions, etc. We should be prepared to deal with the additional routing and provisioning complexity involved in handling additional customers doing BGP, dual homing, etc., both new customers coming for their second connection and existing customers as they reach out to a second ISP for the redundant connectivity. -george william herbert gherbert@retro.com