Cell Relay Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] Re: Connection and connectionless (was: Rerouting of existing connections)
In article <5r94q3$t6n@mozo.cc.purdue.edu>,
qiwei@ecn.purdue.edu (Qiwei Xiao) wrote:
> Hmmmm... I don't agree with that. Whether a protocol is connection-oriented
> or connectionless is a property of itself. There can be a connectionless
> protocol sitting on top of a connection-oriented protocol and vise versa. So
> the correct saying is that
>
> In layer 4, UDP is connectionless and TCP is connection-oriented. ATM is
also
> connection-oriented if used as a layer 4 protocol.
> In layer 3, IP is connectionless and ATM is connection-oriented if used as
> a layer 3 protocol.
> In layer 2, Ethernet is connectionless and ATM is connection-oriented if
used
> as a layer 2 protocol. (I believe there is a HDLC operation
mode
> which makes Ethernet a connection-oriented protocol)
>
> Therefore, the question of whether TCP/IP is connection-oriented or
> connectionless is itself not a good one.
Actually, I have to agree with your points, Qiwei. My position on this
would be, though, that the only "connection-oriented" stuff ATM does is
stuff required to set up a circuit. It's not between applications, at
least not typically. Since these handshakes are used to set up a circuit,
might as well call it that.
In addition, it's hard for me to accept the designation "connection
oriented" when the ATM net consists of nothing more than a mesh of PVCs.
Yes, the PVCs are circuits, but it's really stretching credulity to call
such a net "connection oriented." Whereas its obvious that such a mesh of
PVCs belong to a circuit-switched network.
Bert
manfredi@arl.bna.boeing.com
-------------------==== Posted via Deja News ====-----------------------
http://www.dejanews.com/ Search, Read, Post to Usenet
|
|