Cell Relay Archive

[vhalkka@cc.helsinki.fi]

In article <9807210220.AA11956@noya.bupt.edu.cn>,
Yu Ning  <yuning@mindless.com> wrote:
>Hi,
>
>>There are just more alternatives to use the SONET/SDH infrastructure. Of
>>course there is no reason to  have anything too complicated like ATM
>>cell-slicing on a point-to-point SDH link between two routers, but there is
>>also no reason to force everything to go through a router, if ATM can do
>^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>But the situation is, today 70% of traffic is of IP, then we can't live
>without routers,:-) Even with ATM, its primary means is to connect IP
>routers, whether full meshed PVP/PVC, or SVC.

If the network logical topology is about the same as physical topology, a router
network is OK. We discarded FDDI -based router network because this started
to be a problem. Now we have several OC3 connected LANS that span large physical
areas without a router in between, the whole country if we *really* need it, and the
routers are just somewhere in the net, in convenient places. 

The backbone structure carrying these separate LANs is a redundant loadbalancing
low-latency NET, not a (spanning)TREE like in any shared LAN solution.

I would really hate to replace my ATM switches with equal amount on Cisco 12000
routers, and any IP routing protocol that tries to give same load balancing and
redundancy options..

The ATM/POS solutions just don't compare, at least yet.

VesA