[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [DNA-BOF] Replacement term for L3 Link?
Hi Greg
Kindly fine my in line comments.
[omitted]
> > As mentioned, 'link' is an entrenched term, for example, in 'link-local scope' or
> > 'on-link prefix'.
>
> Though it's entrenched it's not a great term.
May not a great one but not an awful term either. It doesn't seem to me that
the term is so awful that we have to attack an entrenched position right now. :-)
> I don't think it's too much trouble,
> I'm just worried about the time to get
> something discussed and agreed to.
>
> For something short and relatively constrained,
> like the charter, we may even be able to
> use L3 Link as a term (although we depreciate it).
>
> For the development of documents, though,
> there should not be any problem with defining a
> new term so long as it's used consistently, and
> well explained.
>
> For example, the term 'Label' was adopted for MPLS,
> and now everyone knows what that means (even though
> 'Tag' was previously in fairly wide usage).
The new term, for example 'IP Neighborhood', means the same as 'link' defined in
RFC 2461, right?
Because the new term will be used not only inside DNA WG but also OUTSIDE
of it, if we use two different terms for the same thing, that may cause confusion,
unless the new term completely replace the term 'link'.
Moreover if we change the term, we have to change a lot more. For example,
currently we use the term 'Link Identifier'. If we change 'link' to 'IP Neighborhood',
is it O.K to keep using the term 'Link Identifier'? Or do we have to change it to
'IP Neighborhood Identifier' too?
For me, the inherent ambiguity in the term 'link' is not so great for all those troubles.
> > Allow me to delve into detail a little and kindly find the following example.
> >
> > Assume two 802.11 b APs are one the same ethernet link and share one
> > AR like below.
> >
> > |
> > __|____
> > | |
> > | AR |
> > |___ __|
> > __________|__________________ _____
> > | |
> > __|__ __ |__
> > | AP1 | | AP2 |
> > |____ | |___ _|
> >
> > In the above example, there are three L2 links, one ethernet link and two 802.11 b
> > links. But they are connected with bridges(AP) to form one link. Any two nodes
> > attached to the APs can communicated at the link layer.
> >
> > But if a node moves from AP1 to AP2, there is a L2 link change. It should change
> > its AP association and BSS ID.
> >
> > Hence when a node moves from AP1 to AP2, there is a L2 link change but not
> > a link change.
> >
> > A little tricky but I can live with it. :-)
>
> That example was even too complicated for me.
Though complicated, we can't avoid the example. That's the way how currently 802.11 b
APs are deployed.
And for complexity, you should have seen the example I had considered but decided not
to post. :-) In it, two routers share an ethernet link with two access points and the routers
advertise two different prefixes. So in one link, there are three L2 links and two subnets.
> I think it's worth avoiding terms which require
> even the chairs of group to read carefully.
To avoid confusion, we don't have to define a new term. We only have to put either 'L2'
or 'L3' before 'link' accordingly.
> The question is:
>
> Do we change the charter now, to a newly defined
> term, or do we leave the charter alone and wait
> until we all agree to a term which doesn't have
> prior (misleading) connotations?
IMO, Leave the charter alone and if there is a consensus to adopt a new term, change
it then.
Best Regards
JinHyeock Choi