It is one thing to design a network for something which takes 30% of traffic. It is quite another to add new layer of complexity for something which is going to be less than 1% of traffic in few years. Native IP over fiber is not a religion; it's a solid engineering. And you never _have to_ have synchronous circuits to provide service to end users. The brainwashed IT managers who equate voice with synchronous transmission or resource reservation are entirely different breed. Of course, the history is full of examples of people getting rich selling snake oil. Unfortunately that business never lasts long. OSI and VLAN yesterday, CBR today, multicast and IPv6 tomorrow. Right, yeah. I'm not try to tell that there aren't any broken-as-designed legacy junk which one has to deal with time to time; but doing any sensible design for the future means that some old stuff just not going to work anymore. --vadim Stan Hanks <stan@networkmercenaries.com> wrote:

As great as POS is, and as exciting as running IP on optical OC-48 onto WDM gear sounds (no piece of cake, believe me...), there is STILL room for ATM. More than just the much-touted network traffic engineering value, too.

Simply put, because of the B-ISDN heritage of ATM, I can actually run synchronous circuits over it (using circuit emulation) and have it work right. Can't get there from here with any IP-over-glass solutions, alas.

And as much as an "IP over everything" bigot as I've been for the last 17 years or so, I still run into situations where what I *HAVE* to have is some kind of synchronous circuit. Wish it weren't so, but I can't just tell people "oh, you can't do that"...