Cell Relay Archive

[manfredi@arl.bna.boeing.com]

In article <33D75E53.2D8@tcom.epfl.ch>,
  Jean-Philippe Martin-Flatin <jpmf@tcom.epfl.ch> wrote:

[ ... ]

> > At least, that's not the normal way of doing business.
> > (I suppose one could implement a special-purpose ATM network
> > that way, to somehow force VCs to be torn down and re-established
> > whenever a routing table change occurs.)
>
> OK, this answers my question: you think rerouting existing long-lived
> VCCs could be done, but nobody does it. Right ?

Others have mentioned proprietary solutions, like "soft PVCs." But
essentially, you're right.

> So we would need a layer above Q.2931 to first request a new VCC between
> the same end nodes; Q.2931 would then create it based on the new routing
> tables; then the layer above would switch the traffic from the initial
> VCC to the new one, by updating lookup tables in the 2 end nodes;
> finally, the layer above would ask the Q.2931 layer to tear down the old
> VCC. Right ? What would this "layer above" be ?

This might be more complex than it needs to be. The schemes I would use
would perhaps have two criteria that end users would go by:

1. Have communications to the other end user degraded? If so, tear down
the VC and set up another one.

2. For long lived VCs, you might also establish a period of time after
which the existing VC is torn down and a new one established. This would
avoid situations in which the long-lived route has become hopelessly
inefficient.

Using 1 and 2, the end users need not worry about routing tables. You
don't want end users involved in PNNI details, right?

> > This is what makes circuit-switched and packet-switched systems
> > different (ATM being circuit-switched).
>
> Pardon ? Could you please elaborate on this ? For me:
>
> - telephony is connection-oriented and circuit-switched
> - ATM is connection-oriented and packet-switched
> - IP is connectionless and packet-switched
>
> The big conceptual innovation in ATM came precisely from disconnecting
> the concepts of "circuit" and "connection", as far as I know.

Hmmm. I don't think the concepts of "circuit" and "connection" were ever
the same. For example, TCP/IP was always "connection oriented," and yet
no one ever accused it of being based on circuit switching.

What I think the innovation of ATM was is that it made a circuit-switched
system _not_ necessarily constant bit rate. Not tied to the physical
layer's clock, necessarily. In other words, ATM allowed you to have a
circuit-switched network that would not tie up a fixed amount of
bandwidth for a given connection, necessarily. This was not possible for
previous circuit-switched networks. So with ATM, you can choose just how
much of the stable data path benefits you want to pay for.

For example, ISDN gives you a circuit switched connection, but you're
always tieing up n*64 Kb/s of bandwidth, for as long as the connection is
made.

On your last bullet, whether IP is connectionless or connection oriented
depends only on the transport protocol (Layer 4) you use, not on the
underlying network technology.

So, for example, UDP/IP over ATM is connectionless,
                 UDP/IP over Ethernet is connectionless,
                 TCP/IP over ATM is connection oriented,
                 TCP/IP over Ethernet is connection oriented.

Bert
manfredi@arl.bna.boeing.com

-------------------==== Posted via Deja News ====-----------------------
      http://www.dejanews.com/     Search, Read, Post to Usenet