pjw@usna.navy.mil, , jw@math30, (Peter J. Welcher (math FACULTY)) (05/15/91)
I've been reading this and other lists for quite a while and haven't seen a discussion of the following: Portable Unix boxes are well on their way. They tend to have Ethernet transceivers. These can easily plug into networks on more than one subnet, typically. How do people handle assignment of Internet addresses ? It seems that one would really like some sort of dynamic host name <--> address mapping, so as to not have to duplicate entries across subnets (i.e. host foo has an address on each of several subnets, depending on which one its plugged into today). My specific concern is that I administer 2 subnets spanning a college building with about 100 faculty and 40 classrooms. Moving Suns around at the end of every semester is do-able but not fun. As computing power and bandwidth requirements go up, I can conceive of my building being split into smaller and smaller physical subnets / segments, increasing the fun of moving machines between offices. Moving portables as the urge strikes people to relocate could be a real pain. The ultimate nightmare might be someone who wants to take his/hers to class in each of a couple of rooms, and then back to their office, all over the span of a couple of hours. If all those locations aren't on the same Internet subnet, what happens ? Am I missing something ? I seem to be assuming that physical subnet corresponds to address subnet. Yes, you can segment with brouters to cut contention, go to FDDI backbone connecting subnets. But how do you deal with people who can plug into different branches of the tree ? What solutions are on the horizon ? What net architectures address this ?