internet subnets vs bridged ethernets

[Tim Seaver]

Has anyone run into the problem of running multiple subnets
of a single class B internet network on a single (bridged)
ethernet? We're looking into it, but there seems to be no
way (in 4.3 BSD or Ultrix 1.2) to gateway through any host
on the ethernet that isn't also in the local subnet, because
the local host does not have an interface on the remote subnet
(which is really local), as its ethernet board is configured
with the local subnet address.
(Repeat that three times, quickly...)

We can't do without the subnets because we really do have other
subnets gatewayed into the bridged ethernet. The obvious answer
is to make the bridged ethernet a single subnet, which is what we
may end up doing, but the ethernet will expand slowly, eating
existing subnets as it goes. Going with a single subnet means
changing all addresses in a given subnet as it is eaten by the
ethernet bridge monster. The only other possibility I can come
up with is to allow multiple internet addresses associated with
each network interface, so a host can recognize that it has
direct access to multiple subnets through its single ethernet
board. This might not be a difficult extension to the miltiple
address-family code in 4.3 BSD, but I'd like to hear from anyone
who has faced similar problems or has any good ideas on the subject
before tackling either the single subnet or the 4.3 code. Thanks.

	Tim Seaver
	Microelectronics Center of North Carolina
	mcnc!tas
	tas at mcnc.org