Greg,
It would be hard for me to believe that CSCO would let the upstarts take out what should be their fastest growing segment, the ISP market. Of course this can certainly happen, but I wouldn't bet on it. Hedge maybe..... :)
There is also a huge advantage accrued to CSCO with regards to technical knowledge and experience in network engineering - It is far far better to have CSCO specific on your resume then say NT, at least as I have seen in network engineering jobs. Ability to hire staff who can support your network's growth will also enter management's decisions on which vendor to standardize on and deploy. I had seen this happen in the emerging layer-2 switching market, where CSCO lagged badly in it's technology, but eventually stomped the competition though various advantages. They did not stand still.
I am a novice when it comes to the world of optical networking, so please anyone come and correct me if I misunderstand or misstate:
As I understand it, the current state of the art is to switch light carrier frequencies (lambdas?) without ever entering an electronic device to make switching decisions. This optical-optical switching would be inherently (orders of magnitude?) faster than having to do optical-electronic-optical switching.
However, there is a big difference between switching carriers, switching packets and routing packets.
Switching carriers is akin to what AT&T does with your phone (even though internally they are doing this with electronic packet switching like ATM right now). Routing packets involves running the packets through a computational engine which will determine the next forwarding point towards final destination.
Here is a thought experiment based on the idea that we will have to route packets optically in the future. Perhaps we won't have to route, but I don't see how at this point:
Say we have a largely mechanical form of packet routing & switching infrastructure in place, with routing done by mechanical computers which are inherently slower than the new technology, electronics. The end equipment might be mechanical teletypes, just to complete the picture.
The growth of the new E technology is fantastic because it is far superior to existing mech tech, and people all over the place are scrambling to install the new electronic teletypes and electronic-mechanical-electronic switching and routing systems. It all works pretty fast, but engineers and managers know that all-electronic routing and switching systems would be far superior.
But here is the point of all this: Where is the electronic computational device that will allow electronic-electronic-electronic routing so packets will stay at full electronic speed? In this experiment, electronic computers have not been developed yet. So we would be operating in an environment where you could switch electronic-electronic but could route only by electronic-mechanical-electronic.
In today's world, where is the optical computational device that will allow for light-speed routing of packets to take place? The Internet still runs on packets, and routes need to be computed. This is the hard edge where CSCO holds the cards. We would need to see the Internet go fully optically switched for the routing layer to be obsoleted. I'm not saying this won't happen, but I don't see this yet, and frankly, how this could be done engineering wise.
I would like to see a network diagram which details exactly how such a proposed all optical switched Internet would operate.
hivemind |
