You can't generalise because it's so different metro by metro and country by country. But given that you have this doubling every year, it's important to put in technology that has head room. The post-recessionary technology -- the next sweet spot -- is 10 gigabit, a combination of 10-gigabit Ethernet and OC-192. That's the goal the industry needs to move towards. So component and equipment manufacturers have to deliver breakthrough price and performance. That's what we're concentrating on in our optical division -- driving the cost right down. Wouldn't you see 10-gigabit speeds in long-haul networks that run between cities rather than within the metro?
The desktop is going to move very quickly to requiring 1-gigabit capacity. So imagine a building where you've got hundreds of PCs all generating 1 gig in traffic. What you do is aggregate them with a 10-gig link (to an Internet connection). And so, 10 gig is going to become a big issue in the enterprise in the next two years. You'll see it in the enterprise and the metro space, and the combination of both those traffic levels going up means that people are going to have to switch on more (dark) fibers in the long haul. That's assuming that they are going to fill all the excess capacity in the long haul. But to use 1 gigabit of traffic wouldn't you need to have audio over the Internet and video conferencing as a fact of life? I think it's safe to say that that hasn't arrived yet.
That hasn't arrived, and it's not likely to arrive. But you don't need it. What you need in order to need 1 gig are fat e-mail files as well as remote database access. The reality for all of us is that our e-mail attachments are getting bigger and bigger. You might have to wait a couple of minutes for your e-mail to come in over the network. Most people spend a lot of time synchronising their e-mail back and forth. The average knowledge worker is permanently doing e-mail, and gigabit Ethernet makes an enormous difference with that. Gigabit Ethernet also means that almost for the first time you now have the same speed for access across the LAN as you do (to get into) your hard drive. More and more (businesses) are centralising their applications onto servers, and you can now access them as if it's local. To get to some Intel-specific questions, which of your network processor competitors are you watching the most right now?
The market had somewhere between 20 to 30 vendors, and right now it appears that there are only four or five who seem to be in it for anything like the long-term. I think that in the network processing space, serious players have their own fabs. The ability to provide microprocessors over a 10 to 15 year period reliably to customers without your own factories has proven over and over again to be a flawed strategy, because you end up with capacity shortages (and) then you can't supply. So to us, the serious players are going to be the ones who will put the money on the table to build their own factories. Intel started making chips for the lower end of the market -- or the slower speeds -- and the company's recent 10-gig chip announcement signifies that it's going after the high end of the market. Would you say that Intel is going after the incumbents in that market?
The higher-end market is very much up for grabs, because so far (equipment makers) have used ASICs (chips) in that market. They haven't used network processors, (which are more suited to the job.) It's really only been in the last six months that the people who make these high-end systems are taking a very serious look at network processors. The decisions are likely to all get made inside the next 12 months, (so) it's really crunch time for the high end of the network processor market.
Have your say instantly in the Tech Update forum.
Find out what's where in the new Tech Update with our Guided Tour.
Let the editors know what you think in the Mailroom.
