[RAM] revised draft proposed definitions

[RJ Atkinson <rja at extremenetworks.com>]

	Here is a revised draft proposal for terminology,
with the goal being greater clarity of communication.  This
is still not fully cooked yet.  No terminology will be perfect;
the goal here is to come up with something that most folks can
tolerate and that is "good enough" to clarify our discussions
(here and in related fora such as the IRTF Routing RG).

	Thanks to Dino, Scott, and Joel for their offlist feedback,
some of which is reflected below.

	One good question raised off-list is how to classify
an IEEE MAC in this terminology.  On the one hand, an IEEE MAC
does not really have any location embedded inside it.  IEEE MAC
values belong to the end system interface and remain constant
even if the interface's point of attachment changes (i.e. the end
system moves location), so they could be called an Identifier.
An IEEE MAC contains several items of information [Manufacturer
prefix + scope bit (global/local) + multicast/unicast bit +
end system value].  The "end system value" indirectly might imply
when the system was manufactured, but is otherwise entirely opaque.
On the other hand, the IEEE MAC is used directly (feed a Destination
MAC  into a table and get back which I/O port(s) to transmit
it upon) to bridge frames within a layer-2 infrastructure.

Yours,

Ran




Identifier:	An object that is used only for identification,
		never for forwarding packets or determining location.

		Identifiers might exist at different protocol layers.
		For example, a fully-qualified domain name is one
		sort of Identifier.  The EUI-64 in the low-order
		bits of an IPv6 address might be an Identifier,
		for example in an proposed protocol of the "8+8" class,
		though that would be a different kind of identifier
		than an FQDN.

ID:		Abbreviation for Identifier.  See "Identifier" entry.

Locator:	An object that is used only for forwarding packets
		or determining location, never for identification.

Address:	An object that with mixed semantics, where it is
		sometimes used for identity and sometimes used
		for packet forwarding.  Examples include, but are
		not limited to, IP addresses, which are sometimes
		used to forward packets and sometimes used for
		node identity (e.g. in TCP session state).

LISP RLOC:	An object used for forwarding packets in the
		LISP protocol proposal.  This is typically for
		forwarding outside of an edge site.

LISP EID:	An object that has scoped location semantics
		(e.g. used for location when inside an edge site)
		and has end-to-end identity semantics.
		Since it has mixed semantics, this is a kind of
		address, rather than being a pure identifier.

Scoped Locator:	A locator that has non-global scope.  Note that
		a scoped locator only has location semantics,
		never identification semantics.

Scoped Identifier:  An identifier with non-global scope.  Note that
		a scoped identifier only has identity semantics,
		never location semantics.

Identity:	A possible property of an object.  Addresses and
		Identifiers are examples of objects that have
		identity semantics.  Identity can exist at multiple
		protocol layers.  One might compose a transport-
		layer identity from an IP address and a TCP port
		number or from a node identifier and a TCP port
		number -- for example.

Location:	A possible property of an object.  Addresses and
		Locators are examples of objects that have
		location semantics.   Location also can exist
		at multiple protocol layers; most notably,
		a node typically has both a link-layer location
		and a congruent but distinct network-layer location
		at a given instant in time.

Identifier/Locator split:	A class of network protocol that
		has no addresses, and only has (pure) identifiers
		and (pure) locators.  Proposals in the GSE/8+8
		class of solution might be examples of this,
		depending on the details of the proposal.

Multi-Address split:		A class of network protocol where
		one type of address is used for forwarding in one
		portion of the internetwork and a different type
		of address is used for forwarding in a different
		portion of the internetwork.  Simple NAT is not
		an example of a multi-address split, since the
		same *type* of address (IPv4 xor IPv6) is used
		in all routing domains.  LISP might be an example
		of a Multi-Address split.

EOF

_______________________________________________
RAM mailing list
RAM at iab.org
https://www1.ietf.org/mailman/listinfo/ram