NetNews Usenet Archive 1992 #16/NN_1992_16.iso/spool/comp/dcom/cellrel/233

Internet Message Format | 1992-07-22 | 3.4 KB

Path: sparky!uunet!dtix!darwin.sura.net!mips!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!usc!isi.edu!finn
From: [e-mail] (Greg Finn)
Newsgroups: comp.dcom.cell-relay
Subject: Re: Packet Sizes
Message-ID: <[e-mail]>
Date: 22 Jul 92 23:48:20 GMT
References: <[e-mail]>
Sender: [e-mail]
Reply-To: [e-mail] (Greg Finn)
Distribution: na
Organization: USC-Information Sciences Institute
Lines: 61

In article <[e-mail]> [e-mail] (Rajiv Dighe) writes:

>Historical Overview :-)

>Unfortunately, it was widely believed that fixed length cells
>were easier to switch than multiple length and so a standard was
>decided based on the design of a fabric.
>That the cost of the network was not in the fabric but in the OA&M
>was not considered.

>Our arguement then that this fixed-length decision would only add to
>the complexity at the end-points due to the segmentation and reassembly
>is being validated if you look at the number of different adaptation
>layers we already have.
>
>Rajiv-looking-back

    No argument there.  Those who design the CO-to-CO and
long-haul switching systems look at communication from their points of
view.  We are discussing the utility of ATM as a LAN link-layer
technology.  You have pointed out why ATM is not particularly well
designed from the point of view of computer networking ... extra
end-point complexity.

    Some of us in the networking community might not like the PTTs
particular choice of technology for their long-haul communication
networks, but it is generally not our area of expertise, and non-LAN
communication is not the question here.  ATM proponents push ATM as an
end-to-end solution.  Now expertise is not an issue.  Rather the
opposite it seems to some of us, since ATM technology was certainly
not designed as an efficient LAN technology, but rather as an
efficient isochronous switching technology within which asynchronous
traffic can be hosted.

    When using ATM in LANs, all the end points do indeed need to
pay that additional cost in complexity that you mention and also the
lower channel utilization that results from fixed versus variable
length cells.  To the extent that most LAN sourced traffic stays
within LANs today, it seems silly to require every end user to pay
those costs.  An ATM-gateway could perform this function itself, the
added complexity being amortized across a set of users and for just
the traffic that requires it.

    Non-ATM LAN technologies promise to be at least as fast,
require much less host interface hardware, have lower interface cost
and support variable length packets.  The only thing that they seem
not capable of doing more efficiently is interfacing directly to the
longer-haul MANs and WANs.  Does that one advantage outweigh the
others I just stated.  I think not unless the majority of LAN traffic
crosses wider-area ATM networks.

    The decrease in interface complexity makes possible a smaller
interface.  Message routing technology already exists in labs that
allow microprocessors to interface directly to 600+ Mb/s duplex
point-to-point channels.  I think we are poised for another
generational leap in networking scale.  I do not worry about the cost
of the "network" interface but rather about the cost of the
multi-meter cabling technology.
--
Gregory Finn    [phone]
Information Sciences Institute, Marina Del Rey, CA 90292