Re: names, addresses, routes

On woensdag, sep 3, 2003, at 17:33 Europe/Amsterdam, Dave Crocker wrote:

> PN> Well, I consider an *identifier* as something that is more or
> PN> less intrisically bound to an identity and a *name* as something
> PN> that merely indicates an entity,
> I had not run into this distinction before.  Now that you raise the
> point, I guess I have been thinking of identifier as an intentionally
> fuzzy, general term, with name being a more specific reference.
Funny, I would assume the opposite. I'm sure there are truckloads of 
people who have the name "John Smith" in their passport (although I 
have no idea what parents called Smith are thinking when they name 
their poor offspring John), but unless there is a serious screwup 
somewhere, each passport uniquely identifies the bearer, traditionally 
using such distinguishing properties such as date and place of birth, 
height, hair and eye color and so on. I assume that these days everyone 
has their country's take on the social security number in there too.
When we translate this to IP, we notice that there are indeed many 
systems with identical canonical names, but the domain name system 
comes to the rescue and provides us with a naming hierarchy that makes 
it possible to turn locally used canonical names into globally unique 
names. However, it's important to note that an FQDN often isn't the 
"real" name of that which is referred to, but rather a compromise 
between the properties that make for good names in the real world and 
on the net, and, of course, availability of the desired name.
I don't think we have any real identifiers in IP. If there is something 
you want to identify, it's usually possible to find some handle to do 
this. A host can be identified by its fully qualified domain name or 
its attachment to the network, for lack of a property that really 
identifies a host, if something is even possible. (Think about it: 
which part constitutes the essence of a host, the part that you can't 
take away without changing its identity?) But all of this is subject to 
change.
So we should probably be pragmatic and use existing identifying 
properties if they are usable, or create new ones where necessary. It 
seems silly to identify a host by the mathematical property of the 
contents of a small file on the file system, but it works very well for 
SSH. The same thing for the highest or lowest IP address that a box 
has, but BGP and OSPF seem to thrive on it.
FQDNs and IP addresses make more sense but lack a property that is 
sometimes important: they identify something, but don't don't bring 
enough information about the nature of what they identify with them. 
Does a FQDN point to a host or a service? Does an IP address point to a 
point of attachment or a host? Usually we don't care, but sometimes we 
do.
> Addresses contain information about location -- typically topological
> location. (The possibility of geographic location gets bandies about,
> sometimes.) Addresses are also globally unique.
> Routes give directions about how to reach the entity. Routes are
> relative to the starting point.
Everything is relative to a starting point. "Globally unique" just 
means we agree on the starting point.
> PN> In other words, an identifier implies sameness and stability of
> PN> the identified entity,
Obviously the stability of the identifier is limited by that of the 
entity being identified. And an identifier is what you make it.
> PN> while a name does not have those connotations to the same extent.
Yes, names can change, but that doesn't disqualify them as identifiers 
automatically.
> PN>  From this point of view, IP addresses are identifiers.
So what do they identify for what purpose?

I have no trouble identifying my webserver by its IP address, which 
hasn't changed in three years. Not so much with my laptop, as its 
address changes several times a day.
> PN> However, they
> PN> are not end-point identifiers but identifiers for topological
> PN> locations within the routed network.
In other words, addresses are addresses.

> In trying to think about mobility, I have been finding this point
> extremely helpful. IP addresses define a point of attachment -- a
> network interface -- rather than a host interface, as we are used to
> saying.
Do they? What is the point of attachment? The subnet or the individual 
address? I don't think it can be the latter (within this context) as 
the interface brings as much to the table as "the network" in this 
case, in IPv6 at least. (64 bit prefix learned from a router + EUI-64. 
Guess what the I in EUI stands for, BTW.)
> It's worth noting that client-side mobility is a very different problem
> from peer-to-peer mobility.  Hence, something like MAST is entirely
> adequate when it is only the client that is moving around.  The client
> can re-find the server the same way it originally did.
Very good point.