Thanks, Paul.
I've read this article before, and I recognize the merit [fire power] behind such an architecture, but I also understand the implied limitations, ultimately, of its dependencies. Maybe we should not limit this to any specific company, but the principles involved, in general.
For those organizations who can afford to isolate themselves, from a connectivity standpoint, and for whatever reason don't care if they are, or are not, universally "connected" from every end point of such a network, then a modified ATM (with or without IP) template can be made to work out fine.
On the other hand, if an organization wants to truly integrate and leverage its infrastructure to meet all possible forms of networking contingencies, then I have some reservations about taking a path that is heavily dependent on ATM, or any proprietary implementation of hybrid technologies that bound enterprises to a single networking texture. There are reasons why one would want to do this, of course, and there are reasons why this would be unattractive. I would have to look at this some more, before forming a more conclusive opinion on this direction.
While it's true that no network need truly be an island unto itself (you could always use additional routers, gateways, and bridges to get to the "outside world," right?), it is also true that one-offs and proprietary implementations that buck the tide of what everyone else is doing are more difficult to administer in the end, and wind up costing potentially a lot more (total cost of ownership) than they save... if universal connectivity is an objective.
Here's a link that may spell out some of the types of concerns I would have using technologies that limit connectivity to a closed group.
useit.com
Regards, Frank Coluccio |
