RE: stable addressing

["Fleischman, Eric" <eric.fleischman@boeing.com>]

>Not that this question is really on-topic for multi6 but it would be 
>interesting to better understand what the requirements for companies 
>that insist on their own address space are. "Not being hostage by the 
>ISP" is one argument, but that often boils down to readdressing scare. 
>However, with many of the id/loc ideas that have been presented here, 
>readressing is becomes another issue to what we are used to today. 
>Anyone like to bite on this? Perhaps we should take this off-list 
>though.

Kurtis,

There are doubtlessly many metrics that can be used to determine whether a corporation or government should own its own IP addresses or whether they should be leased from some ISP. Some have suggested that this could be based on the address block size (e.g. /32 or /48 or some other number). Another metric could be business based: what is the size of an organization such that an ISP would be willing to contractually handle the routing to that corporation using IP addresses from another ISP?

Concerning readdressing, only a very small part of the problem with readdressing is the actual changing of a computer's address. The much larger problem is finding and modifying all of the other systems which are keyed to the old address. DDNS does this for DNS for dynamically addressed systems. However, most key services are statically addressed. Yet this misses the true point: how about the proxies, firewalls, gateways, and many other systems which may also need to be modified? As the IP Layer becomes fatter, the problem with readdressing grows substantially. 

Also, the identity problem again rears its ugly head here. If an entity has been so short sighted as to use IP addresses as computer identifiers, then a whole constellation of databases and directories will need to be modified.

--Eric