I apologise that I did not organise my thoughts well earlier,
so I will try to complete -- hopefully in a more clear way
-- the thoughts I was trying to express earlier in the WG mtg.
TERMINOLOGY;
Anyplace I say "home" or "residence" below, I also
mean to include "small business" or any other smallish
organisation that might have a HomeNet type deployment.
OBSERVATION:
The HomeNet WG has at least 2 distinct communities
in its membership, with at least 2 sets of objectives.
There might well be more than 2, but hallway discussions
this week make it clear to me that we have at least 2.
So I think it unlikely that the WG can converge on
any single solution for unicast routing.
I doubt it is worth a lot of effort to try to force
a single solution. I am not even sure it is necessary.
USER COMMUNITIES:
The WG has at least 2, possibly more, communities
within its membership. These have at least some
divergence in their goals/objectives, although there
are some areas of commonality.
A) Service Providers
Several service providers want a Homet Net solution
that lets them manage/provide/trouble-shoot the home
networks inside their residential customers' homes and
apartments. Service providers are comfortable with
link-state routing and probably prefer IS-IS over OSPF
for cultural reasons. Their goal is to reduce their
operating expenses by reducing/eliminating telephone
calls from users.
So they will prefer a link-state approach to unicast
routing within the home (or small business). This community
is quite happy to adopt DHCP options and other IETF technology.
In many cases, but not all cases, the equipment budgets
are less critical than the potential for operating cost
reductions (because operating costs recur, while capital
costs are periodic). Something like HNCP has a good chance
of also being included in their solutions.
B) Consumer Electronics Vendors
Many home users want to purchase their own in-home
networking equipment and do NOT want to use provider-
supplied equipment. The same is true for many small
businesses. Consumer electronics vendors already
are quite happy supplying this market. Consumer electronics
vendors operate with thinner margins (e.g. eliminating one
resistor makes a meaningful difference in their profit).
Memory footprint and CPU cost matter a great deal to these
manufacturers. They already have RIP, RIP is nearly zeroconf,
and adding a link-state routing protocol would increase
both their manufacturing cost and their support cost.
Moreover, RIP scales just fine for residential and even
small-business networks. It is nearly zeroconf, which
is a bonus.
So these folks are not likely to add any link-state
routing protocol to their products. They are likely
to add a few DHCP(v6) options that are useful and also
are likely to add at least parts of HNCP (maybe the whole
of HNCP if the footprint/complexity is small/low enough).
PERSPECTIVE:
It would optimise WG productivity to focus on matters
other than trying to select a single unicast routing protocol.
We can focus on making HNCP as good as it can be, identifying
relevant DHCP(v6) options (possibly new options), identifying
outputs from the IETF DNS-SD work that are applicable
(and citing them as appropriate), and sorting out the other
non-unicast-routing parts of the Home Net problem space.
Yours,
Ran
PS: I agree with whomever spoke up earlier today about the
importance of supporting IP multicasting throughout a
HomeNet deployment.
_______________________________________________
homenet mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/homenet
