Up with IP

The March 31 Perspectives (page 68) was largely well-written and balanced, but the quotations seemed to tell only one side of the rather charged IP-over-ATM debate.

In particular, a rather dubious suggestion by Tom McDermott III of Alcatel that IP switches/routers could not sustain the additional traffic resulting from even a 4% increase in ADSL penetration went wholly unchallenged. McDermott's statement that ATM is a transport mechanism to bring all IP to one place to deal with it is precisely the fatal flaw in many engineers' understanding of IP-over-ATM in a wide area environment.

Unlike many data networking protocols more familiar to many in the telephony community, IP does not use virtual circuits. IP is fundamentally a packet-switched protocol like ATM. Even if you get all your IP traffic to one place, you still have to route it. Bringing it to one place gives you the additional work of bringing it all there in the first place!

Furthermore, there is vastly more experience in the field running interconnected, survivable, high-performance TCP/IP networks - which "switch" IP, though it is generally called "routing" - than there is running high-capacity ATM networks handling primarily data. And though it is not surprising that many equipment vendors with a stake in ATM switching technology heartily recommend it, the fundamental inefficiency in running one dynamically routed, packet-switched protocol layered on top of another is fairly obvious. As the explosive growth of the Internet from its humble origins demonstrates, IP is a very good packet-switched protocol for data.

Finally, it is a common fallacy promoted by many ATM switch vendors that TCP/IP routers are somehow less scalable than ATM switches. IP can be switched in hardware just like ATM can, not only in a LAN environment but in a wide area environment as well. In fact, Cisco Systems deployed fast-path packet switching for IP many years ago (the Silicon Switch Processor on the Cisco AGS+ router) and along with others, has hardly stood still in the interim. Actually, because ATM cells are far smaller than typical IP packets, one need switch far more of them per second in order to carry the same amount of data.

ATM equipment vendors would appear to have the technical issue precisely backwards: ATM switches are less scalable than IP routers.

Thor Lancelot Simon U.Communications, LLC

No guarantees The April 14 Perspectives (page 64) hit it right on the head. To listen to the CNBC (or any financial analysts) talk about the market, you'd think there were laws against a bear market or any loss.

Stock values should somehow be incorporated in the CPI. How can the value of a company be 'inflated' without affecting the economy?

There is an overemphasis on the stock market, and it will come back to bite us in the @ss! Jim Ford WebTV

In a day and age when people, from the top on down, refuse to take responsibility for their own actions, I can't tell you how refreshing it is see someone, especially a member of the press, advocating personal responsibility, as you did.

As an employee and shareholder of a large telecom company, I too have been frustrated at stock declines. I would never consider blaming senior management for my personal loss - it was my decision to buy the stock. Our company has far exceeded its goals year after year, and still the stock price does not budge. Such is life! I knew the stock was risky when I bought it.

And yet, all around me, I see people unwilling to take the bad that comes with life. They always blame someone else. Thank you for being bold enough to stand for the unpopular but right position that you stated in your article.

Michael Welter