The real cost of bandwidth - network management challenges - Industry Trend or Event - Column

Communications News, June, 1999 by Lenny Liebmann

It's not just the raw dollars-per-Mbps that makes networking expensive.

To hear some people talk, you'd think that bandwidth was like gasoline: when prices fall, you just pull up to the pumps and fill 'er up. Now, it's true that large-scale service-provider network buildouts, new approaches to networking (such as Internet VPNs), and high-bandwidth technologies (such as DSL and gigabit Ethernet) have created much greater Mbps-for-the-buck opportunities. But getting to those savings is not quite as simple as driving down to the corner gas station.

First of all, if I want cheaper gas, I don't have to buy a new car. In the real world of networking, on the other hand, there is almost always a costly--and sometimes even painful--migration process. Think you can convert a global corporation from 16 Mbps token ring to switched 100 Mbps Ethernet overnight? Think again. You have to replace equipment in stages, often maintaining parallel infrastructure over some period of time to prevent interruption of business processes. You have to reconfigure systems and applications. Often, you have to build an entirely new set of vendor/support relationships as well. That all costs money, too.

New networking technologies can also create unforeseen management problems. Flat, switched campus environments, for example, have generated a whole slew of challenges such as IP address administration for which most initial adopters were unprepared. And, if you go for cable or DSL access, you'll find that their inconsistent availability will mean that you have to track and configure more than one type of remote link. That complexity is also a tangible, if somewhat difficult to quantify, cost.

Then there's the dreaded "step function" of bandwidth buying. If you're using a T1 line, your 1.3 and 1.4 Mbps is basically free. But, if you need one more tenth of a Mbps, it will cost you plenty. The same thing is true if you go from a 128-kbps Frame Relay link to 256 kbps. You don't buy bandwidth by the bit-per-second (at least not yet). You buy it in specific packages. When you max out that package, you usually have to graduate to a significantly more expensive and robust technology--even if you only need an incremental increase in capacity.

A similar principle holds true in campus environments. Because of the interdependencies of data-communications architectures, it's often very difficult to perform incremental, piecemeal upgrades of local area networks and achieve any real performance gains. What's the point of setting up a switched workgroup if you still have only one segment leading into the server? So, to avoid simply pushing the bottleneck around, it's usually necessary to do a lot more than just stick one faster piece of hardware into the situation.

And whoever said that equipment and service-provider charges were the biggest cost constraint for network expansion? Isn't there also a shortage of qualified technical staff? It doesn't matter if you have all the money in the world to spend on switches and routers; if you don't have someone who can evaluate, design, and deploy a next-generation network, you're going to remain stuck with what you have.

Those human resources are closely linked to the most limited resource constraint--a resource that no corporation anywhere has ever been able to purchase more of at any price: time. Despite all our advances in technology, no one has successfully engineered a 30-hpd (hours-per-day) chronology solution. Many of the IT managers I interview over the course of a day simply have too much on their plates--including Y2K remediation, Web commerce projects, and organizational change issues--to push ahead with their network upgrade plans. This is often true even when a migration to VPN technology could potentially save their company thousands of dollars a month. They simply don't have the personal "bandwidth" to make the decisions that need to be made with the necessary degree of confidence right now.

That's a big part of the appeal of managed services. It isn't that companies don't have the money or can't cost justify network projects. They just don't have the bodies to throw at the problem. So they actually part with more cash, at times, to get the thing done and over with--without having to shoulder any of the ongoing administration and management tasks either.

On top of everything else, bandwidth needs are escalating faster than raw bandwidth prices are dropping. Multimedia content doesn't just need a little more pipe--it needs a whole order of magnitude more.

That's why a clearer understanding of the "big picture" of bandwidth economics is so crucial in today's digital marketplace. Pinning your hopes on falling service-provider costs alone won't do the trick. Successful next-generation networking will require a strategic view of how to apportion capital, human resources, and time. Without that view, your company's e-business initiatives could wind up running out of gas.