Friday, March 21, 2008

SPECIAL REPORT: The Growth of Convergence, part III


There has been a lot of discussion about convergence lately. Based on some recent interest in this information, here is part III of a primer on this trend and what it means to all of us; not just in telecommunications, but in all facets of the business world.

As we discussed earlier, many companies had tried and failed with early forms of convergence. Many of these failures occurred simply because the wrong network was chosen to converge upon. In part II, we looked at AT&T (coaxial cable) and Sprint (ATM) and how these choices decimated nearly destroyed these companies. The choice of today’s integrators is TCP/IP.

Whether or not it is run over the actual Internet or another network, TCP/IP is the language of the Internet. It is superior for the simple reason that it is dumb. Unlike other network types such as ATM and Frame Relay, TCP/IP does not rely on error correction. In other words, so called “Intelligent networks” constantly measure the quality of the information they send insuring QoS (Quality of Service). If a packet of data is not delivered, the network will look for it and deliver it eventually. Sounds good if we’re sending documents, but worthless if you’re trying to make a phone call. The last thing you’d want is the garbled bits of talk that dropped off your call delivered at the end of the sentence. Worse, all that error correction slows the network down.

TCP/IP has no error correction. It is not an intelligent network – quite the opposite. When something is sent and there’s a problem, the message just gets lost. That sounds bad, but its not; As described earlier, if part of the phone conversation falls out, you don’t want it back at the end. It’s the whole concept of dumb cars on smart roads.

Think of the data packets as cars on the road; the road has no intelligence, just the cars (or really, the drivers). If someone is lost, it’s not the road’s responsibility; it’s the cars’. This is a far more efficient system. The same is true of TCP/IP.

Now with TCP/IP as the medium for the converged terrestrial network, anything can be delivered or sent so long as the price is right, and for all the reasons I mentioned earlier, IP is also, cheap. Did I say cheap? It’s getting cheaper all the time.

Fiber-optic cable costs less than copper wire, and the per-megabit rate continues to drop. In 1995, I worked for AT&T and sold my first T1.5 for the shockingly low price of $6,600 per month (that’s 1.44 megabits per second of IP). My fellow sales people were amazed that I was granted a rate of less than $10,000 from the pricing department. By comparison, last month I purchased FIOS from Verizon for my home: 15 megabits per second for $35 per month, delivered by fiber-optic cable.

NEXT WEEK: Convergence in other areas of telco

No comments: