RealTime IT News

We Need Fat Pipes

Reporter's Notebook: Convergence. Triple Play. IP. Ethernet.

After four days at NXTcomm in Chicago, I probably heard each of those words 1,000 times or more. Make no mistake about it, as I had figured before the show even began, NXTcomm 2007 was, indeed, all about convergence, what that means and how to do it.

The big carriers, AT&T , Verizon and Qwest, talked up their convergence and triple plans and all sounded remarkably similar.

The message? Any content on any device anywhere all enabled by IP.

Certainly this is not a new message. What is new is the fact that the carriers now have proof points and big numbers to back up their convergence claims. Verizon now claims to have laid down fiber connectivity of up to 100Mbps to a million homes, and AT&T is ramping up its uVerse broadband connectivity program.

Vendors all over the floor talked the same game of convergence on IP infrastructure. But when you do a deep dive, it becomes apparent that IP infrastructure on Ethernet, though pervasive in the enterprise, is still a long way off from being pervasive in carrier environments.

The telecom carriers, after all, have a century-old legacy of copper in the ground by which the majority of Americans receive their phone service. Cable companies have 50 years of coax that has been laid. And on the backhaul infrastructure, there are layers upon layers of older, non-Ethernet optical technologies to deal with.

There are many different way of enabling IP and Ethernet over the various legacy connectivity technologies that are out there. Carriers have to figure out what works for them as either overlay or new technologies. It certainly is a confusing mess from where I stand.

Putting everything on one core pipe also has its risks. IP infrastructure is fundamentally the same IP approach that is under constant attack on a daily basis. Certainly the carriers will take precautions to secure their networks, but in comparison to legacy cable and telecom networks, the new risks may well be higher.

Then there is the actual issue of bandwidth.

Convergence not only requires a common IP Ethernet-based architecture, it requires boatloads of bandwidth. In session after session I heard the same statistic that carrier network loads are growing at 80 percent to 100 percent every year.

With more video and voice coming onto the network, it's expected the networks will need even more bandwidth. It's bandwidth that is being rolled out, but it's not quite there. Not on the consumer side. Not yet.

As a case in point about the need for more network capacity, I found my way to Chicago's Millennium Park. It's a marvelous outdoor concert facility that draws massive throngs of people that choke network bandwidth. I had some difficulty making a cell phone call (with access through Cingular/AT&T), as the network was apparently too busy. A local Chicagoan told me it's common whenever there is an event.

On the other hand, enterprises are getting increasing amounts of bandwidth made available to them. 10 GbE and OC-192 optical connections provide a good amount of bandwidth.

With Metro Ethernet, the options and opportunity for application convergence for the enterprise are very real. Such vendors as Cisco, Juniper and Alcatel-Lucent among others were keen to talk about the smart network and how the new carrier networks are more application-aware. It's a trend that will only serve to make enterprise computing even more powerful, convergent and possibly even cheaper.

Convergence did blow into the Windy City this past week. It's not a new trend; actually it's not a trend at all. It's the new normal and it's coming to you sooner rather than later.

Sean Michael Kerner is senior editor of internetnews.com.