Internic address allocation policy

Roger Fajman RAF at CU.NIH.GOV
Wed Mar 22 00:40:57 UTC 1995


> I fully agree that host autoconfiguration seems like a reasonable
> technical solution to the problem of changing providers. The importance
> of providing autoconfig for IPv6 *on day one* must be stressed over
> and over again. You, as a member of the IESG, could certain help
> with this.

Host autoconfiguration is not the complete solution for easily
changing providers.  There are many places cuurently where IP
network numbers appear:

   Router configuration files
   DNS configuration files
   DHCP configuration files
   Security filters (TCP wrapper, etc.)
   SNMP configuration files
   NETBIOS over TCP/IP host tables

I'm sure that others can add to this list.  In our organization, at
least, not all of these things fall under the control of those of us
who run the backbone network.  Even when they do, there are still
enough of them to make the job of changing them all non-trivial.  So
implementing host autoconfiguration will be a significant step towards
making it easy to change providers, it is not the complete solution to
the problem.

While it's certainly true that the format of configuration files is not
a protocol issue, it is important that the protocols provide a way of
passing the necessary information around or making it unnecessary for
it to be kept.

Roger Fajman                                   Telephone:  +1 301 402 4265
National Institutes of Health                  BITNET:     RAF at NIHCU
Bethesda, Maryland, USA                        Internet:   RAF at CU.NIH.GOV



More information about the NANOG mailing list