Cell Relay Archive

["Ashish Lal" <lal@alpha.dtix.com>]

I have been following this thread with interest.
What about the post ATM scenario? I agree that currently the ATM and IP
worlds seem to be converging. A few years from now (say about 5 or 10 at the
most), most PCs would have many enhanced multimedia applications running
over ATM and IP. I am wondering what would come next - I mean the post ATM
scenario. A few years back, when there was no ATM, we felt that greater
bandwidth was necessary for quick and easy communication. Does anyone
foresee any further parameters being added in the future that would define
communication over the internet (apart from bandwidth and QoS that ATM
currently provides)? The next technology (say X) that comes after ATM would
certainly improve upon ATM's QoS.
Thank you and warm regards.
--Ashish

>Jonathan Turner wrote:
>> ...
>> With hindsight, it's now easy to see that this was a poor
>> strategy. What's more, it's not hard to see how we might
>> have avoided the trap we fell into. Rather than subordinate
>> ATM to IP, we should have been using IP as a subnet technology
>> for carrying ATM traffic.
>
>Maybe I'm missing the point, but this sounds strange.
>
>I think the real issue is that changing infrastructure is
>(comparatively) easy, while changing end to end protocols
>is very hard.  That's because there are so many boxes
>involved and so many implementations to change, and most of
>them are hard to change platforms like PCs.
>
>So any new networking notion that aims to replace the IP based
>application protocols by a completely new stack at the end
>systems has a VERY large barrier to overcome.
>
>On the other hand, new technology that makes the internet
>backbone work better is a much easier sell.  That's why
>ATM is replacing frame relay, and newer things (Packet over
>Sonet or the like) are emerging as well.  Those changes are
>transparent to end systems.
>
>This is why CIF (cells in frames) failed, as expected.
>It addressed the lesser problem (installation of new NICs)
>without addressing the bigger one (replacement of all the
>applications).  It seems to me that ATM over IP is a slight
>tweak of CIF (moved up a layer) and will have the same
>difficulties.
>
> paul