I think the current terminology is consistent with the client/server
model definitions but still carries a lot of confusion with it.

In the client/server model, the "client" is,as you mention, the one that
"initiates" the connection [read: performs connect()]and the "server"
is the one that accept [read: performs accpet()] the connection request.
So, it is perfectly normal that a piece of software that, for example,
is doing recursive DNS resolution, can be at turns a client and a
server...

The problem is that people get confused by the IT terminology, where PCs
are client and big machines are servers... Or by web-like architecture
where firefox is a web client/browser running on a PC and apache a web
server running on a big iron.

There are several pieces in the DNS architecture, and having a well
defined, commonly accepted terminology for that will really helps.

Those pieces includes (I'm sure I'm missing some):

A) the "things" in the OS of most nodes, a.k.a. the stub resolver/local
resolver/local
name servie caching...Things of course get more complex as soon as
local caching
is involved.

B) the "things" that IT/Service providers set up to resolve DNS names,
a.k.a. DNS cache/DNS forwarders/DNS servers/DNS recursive resolvers.
Those things come in different flavor, perform different roles,
but there is no good name for the generic role they provide.

C) the "things" that DNS administrator set to manage particular DNS
zones,
a.k.a. DNS servers as well.

Of course, that fact that the same piece of software is often use in
those three roleS does not really help to clear the perceived confusion!

It would help a lot in various documents (sometime done outside of the
DNS wgs) if there was an agreed, referencable, terminoly for those three
roles.

- Alain.



> -----Original Message-----
> From: owner-namedroppers@ops.ietf.org=20
> [mailtowner-namedroppers@ops.ietf.org] On Behalf Of Paul Vixie
> Sent: Thursday, February 02, 2006 9:51 AM
> To: Hideshi Enokihara
> Cc: namedroppers@ops.ietf.org
> Subject: Re: [question]What is an accurate meaning of=20
> "client" in RFC2181?=20
>=20
> i believe that the terms "client" and "server" are inadequate=20
> to describe the roles of the various protocol agents in DNS,=20
> and as in RFC 2136, i suggest the terms "initiator" and=20
> "responder" instead. the the parts of RFC 2181 which you've=20
> quoted here, "client" really means "initiator". the DNS=20
> protocol specification must be understood in terms of=20
> messages -- their format and meaning. two messages make up a=20
> transaction, and each transaction is initiated by some=20
> protocol agent and responded to by others.
>=20
> --
> to unsubscribe send a message to=20
> namedroppers-request@ops.ietf.org with the word 'unsubscribe'=20
> in a single line as the message text body.
> archive:
>=20


--
to unsubscribe send a message to namedroppers-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: