Cameron, > > - less IPv4 exit mechanisms to implement and choose among. that must be > > good, no? > > (currently there are at least 4 mechanisms proposed in the A+P H&S and > > mesh space) > > - large stateful boxes in the SP network; we know how to build them, > > and we can charge more for them. that's a good thing too, no? ;-) > > > > Do you have data? Or are you perpetuating the myth that stateless is cheaper?
where did you get the "myth idea" from? cost obviously depends on platform. scaling a solution based on per flow state, versus per subscriber state versus no state at all. it obviously costs money to keep this state around, increased complexity to maintain it and so on. in addition you need solutions for redundancy that typically involves hot-standbys. while a purely stateless solution can easily do load balancing between multiple active systems. of course, if we end up with all platforms have to support all mechanisms, including stateful translation, then these may end up on special service cards anyway... > Last I checked stateless requires some heavy duty provisioning systems and > less efficient static allocation of expensive ipv4 addresses I don't think a DHCP server was considered a "heavy duty provisioning" system. > > > I'm also really tempted to do that if this community can't decide what to > > > do in a quick time manner, showing the industry the path forward. > > > > I ietf has seldom shown the path forward. In this particular context > ....especially so. > > > yes, if there was a path forward I would also be a lot less despondent with > > regards to A+P. cheers, Ole _______________________________________________ Softwires mailing list Softwires@ietf.org https://www.ietf.org/mailman/listinfo/softwires
