Re: What's the good word?

[Fred Baker <[e-mail]>]

>>     Thanks for the response - your use of the term "subsumption" was
>>     what I thought it meant; however it's not defined in any other
>>     RFC or draft that I have seen.  RFC-1058 (RIP-1) also used the
>>     term, but it referred to subsumption as a way of hiding subnet
>>     and host information from routers in other nets only in order
>>     not to break RIP-1.  The way it's used in RIP-2 is as an optimization
>>     rather than a restriction, but the wording almost seems to suggest
>>     that one subnet might be part of a larger one.

	That's *exactly* what subsumption is all about; treating
	separate subnets as though they were a single one when one is
	sufficiently far away that an occasional extra hop is not a
	major issue.  I used the word, I think, the same way that RFC
	1058 uses it.  The difference is that RIP-I uses a different
	set of rules to know when or how to do it, not that it is
	fundamentally trying to do something different.

>>     All I'm asking for is clarification; perhaps some of the text
>>     from your response could be put in the document.

	Gary: do you think you can provide the requested clarification?

>>     As for RIP domains - the wording just doesn't make a lot of
>>     sense.  Why is it "necessary" to allow multiple instances of RIP
>>     on a single wire?

	That is a question you should ask of someone who believes that
	Routing Domains are within the scope of the protocol at hand.
	I think I have made my opinions clear on that subject. :^)

>>     On IP multicasting - you imply that this simply requires the
>>     translation between IP and hardware addresses for appropriate media,
>>     which sounds good to me.  (the only document I have on IP multicasting
>>     is RFC 1112, which says that if you want to send as well as receive
>>     multicast, then you MUST do IGMP)

	May I suggest that you get a copy of RFC 1247 (available in text
	or postscript) and read the paragraphs that mention the multicast
	addresses AllSPFRouters (224.0.0.5) or AllDRouters (224.0.0.6)?

	The relationship between the Ethernet or FDDI MAC Address and
	the IP Multicast address is described in RFC 1112 section 6.4
	"Extensions to an Ethernet Local Network Module".  Steve kind
	of waves his hands at leased lines and other things that aren't
	Ethernet is section 6.5.

	There is one key distinction that needs to be made here.  Steve
	is describing a *WAN-wide multicast* between *hosts*, with
	router support.  IGMP is defined to facilitate that:

****************************************************************************
*APPENDIX I. INTERNET GROUP MANAGEMENT PROTOCOL (IGMP)
* 
*   The Internet Group Management Protocol (IGMP) is used by IP hosts to
                                                  ^^^^^^^^^^^^^^^^^^^
*   report their host group memberships to any immediately-neighboring
*   multicast routers.  IGMP is an asymmetric protocol and is specified
*   here from the point of view of a host, rather than a multicast
*   router.  (IGMP may also be used, symmetrically or asymmetrically,
                   ^^^
*   between multicast routers.  Such use is not specified here.)
****************************************************************************

	Use of IP multicasting in OSPF and RIP-II is far more limited
	in scope: it is by definition between immediately adjacent
	*routers*, and may be listened in on by immediately adjacent
	hosts.  There is, therefore, no need of registration, and no
	WAN-wide list of "all these IP addresses are running this
	routing protocol."

>>     I would also suggest mentioning something about the MIB
>>     somewhere in the protocol document.  (since mandatory counters are
>>     defined there which will require additional processing)

	Maybe.  I thought the MIB document captured what we discussed.

Fred