Cell Relay Archive

[Paul Koning <pkoning@xedia.com>]

S.Ramkumar wrote:
> 
> manfredi@arl.bna.boeing.com wrote:
> 
> >
> > Actually, the issue is not ATM vs TCP at all. It is instead SSCOP vs TCP.
> > Not that I disagree with your emotional outburst, Steinar, but if ATM is
> > to be used for reliable data transmission, it requires a
> > connection-oriented protocol just as IP does. After all, both ATM and IP
> > are best effort media, and it's only misnomer that calls ATM "connection
> > oriented." It is not. It is circuit switched, but still best effort.
> 
> I am thoroughly confused why you said that the term
> "connection oriented" applied to ATM is a misnomer.  To send a cell,
> you need to establish a VCC between the end points.  Doesn't this
> characterize ATM to be a connection-oriented protocol?
> 
> If not, what are the qualities required to call some protocol
> "connection oriented" or not?
> 
> Also I thought ATM was packet (cell rather) switched.
> 
> Kindly clarify,
> -ramkumar

The problem is that there are two independent properties, not just one:

1. State and data routing:
1.a. connectionless: no state, each data unit forwarded independently
(e.g., IP)
1.b. connection oriented: there is state, data unit refers to that
state, 
forwarding is done using information in the state (e.g., ATM, X.25)

2. Reliability of delivery:
2.a. "best effort": each unit of data is subject to being lost without
notice
2.b. "reliable delivery": units of data are all delivered exactly once
and in
sequence OR you get notification of failure

IP is (1a,2a).  X.25 is (1b,2b).  TCP is also (1b,2b).  As a result,
people 
often use the term "connection oriented" to describe the pair (1b,2b)
even
though strictly speaking it refers ONLY to 1b.

ATM is (1b,2a) as is frame relay.

You could refer to ATM as "circuit switched" because it does have
connection
("virtual circuit") state, but I think this is not usually done.  It
certainly
is correct and more common to say that it is cell switched.

By the way, "best effort" (2.a.) services come in several variants: IP
doesn't
guarantee delivery, and it also does not guarantee order.  I think it is
in
theory allowed to duplicate packets though in practice that is not
likely.
ATM, on the other hand (Ethernet and other LANs similarly) is subject to
loss but
is NOT allowed to duplicate or reorder packets.  In practice that
doesn't matter
much because all correctly-designed transport protocols will function
correctly
even if packets are reordered or duplicated (though some will do so at a 
significant performance impact -- if so then reordering should be rare).
There do exist transport protocols that malfunction if presented with
out of
order packets, but that's a sign of a major design defect.

	paul
-- 
!-----------------------------------------------------------------------
! Paul Koning, NI1D, C-24183
! Xedia Corporation, 119 Russell Street, Littleton, MA 01460, USA
! phone: +1 508 952 6000 ext 115, fax: +1 508 952 6090
! email: pkoning@xedia.com
! Pgp:   27 81 A9 73 A6 0B B3 BE 18 A3 BF DD 1A 59 51 75
!-----------------------------------------------------------------------
! "The only purpose for which power can be rightfully exercised over 
!  any member of a civilized community, against his will, is to prevent
!  harm to others.  His own good, either physical or moral, is not
!  a sufficient warrant."    -- John Stuart Mill, "On Liberty" 1859