The feds aren’t normally my first choice for cutting-edge market research. But they have one thing going for them: When they send out a questionnaire, they get responses. The DOC wants to know who is connected to the Internet. The FCC is interested in how quickly broadband access is penetrating the market. Both agencies have some very good news for makers of optical networking gear. Internet usage is increasing in a hurry among all Americans, not just the “haves.” Meanwhile, broadband access is accelerating rapidly, with cable modems in the lead but DSL growing at a faster pace.
But this bull market for broadband also leads to a disturbing question: Are carriers really building the right networks for this Brave New World?
Here are some highlights from the DOC report, “Falling Through the Net: Toward Digital Inclusion” ( http://www.esa.doc.gov/fttn00.pdf).
The FCC’s report, snappily titled “High-Speed Services for Internet Access: Subscribership as of June 30, 2000” ( http://www.fcc.gov/Bureaus/Common_Carrier/Reports/FCC-State_Link/recent.html) quantifies the deployment of broadband access, as of June 2000 (a remarkably recent date, considering this is the FCC we’re talking about).
Is it fair to assume that more users at higher access rates will put significant strain on core networks? Absolutely. Will carriers have to continually scale their optical networks to accommodate this traffic? Without a doubt. And they’ll need more than Sonet and DWDM. This is packet traffic, which will require new levels of intelligence in optical networks and much greater flexibility in responding to Internet apps.
Think of it this way: Consumers — especially children — are spending more time at the computer and less time in front of the TV. Multimedia applications will drive the adoption of broadband access (Napster is already doing just that). That multimedia traffic, in turn, will force the deployment of “high-touch” routers (those that look deep into every packet, ordering them according to many more variables than can a typical core router) at the network edge and massively scaleable switches at the core. Optical networking equipment must not only supply the fattest pipes possible between network nodes but also deal with the variable nature of IP traffic.
Let’s take another look at those government reports. Their main thrust is that the digital divide is starting to narrow, while broadband access is just starting to reach the masses.
What gets overlooked in much of the talk about carrier capex and the alleged softening of the optical market is that the network is still in its infancy. By and large, carrier buildouts are desperate attempts to scale a voice network in a world where data rules. These efforts demand massive infusions of Sonet, DWDM, and cash but do little to cut service provisioning times or increase revenue opportunities. It’s really the old solution of throwing bandwidth at the problem. It didn’t work on the enterprise, and it isn’t going to work on the Internet.
It’s no wonder that carriers are getting frustrated: They’re building the wrong network for the job.
So what’s the right network? There are two answers, and both could be correct.
One camp argues that a dynamically switched circuit network is the way to go, since it leverages the installed base of reliable Sonet equipment. Trouble is, this also means adding the requisite intelligence and switching power to Sonet in order to bring Internet dynamism to a static network of fixed circuits. Sycamore Networks Inc.’s (Nasdaq: SCMR) team-up with Sirocco is one example of this approach in action; Ciena Corp.’s (Nasdaq: CIEN) Core Director is another.
The other camp, only now beginning to flesh out its argument, says it’s time to stop trying to force-fit packet data onto a circuit network. The alternative: Build infrastructure that accommodates circuit traffic on a packet network. ATM was once the poster child for this approach, though provisioning complexities and cost limited its adoption to the network core. Ethernet is now starting to turn up as the new platform on the edge (watch Lantern Communications, Luminous Networks Inc., and Atrica Inc. here), and MPLS is beginning to put a squeeze on ATM at the core (witness Tenor Networks Inc., Vivace Networks, and Caspian Networks).
Carriers that have the luxury of starting with a blank slate are starting to take the pure-packet play seriously, while well-established providers continue to push Sonet into the data world. True, the latest generation of packet prophets claim this is wasted effort, but it’s worth remembering that Sonet was originally intended to transport both circuit and packet traffic. The International Telecommunication Union (ITU)’s original vision of B-ISDN had Sonet supplying the underlying physical layer for multiprotocol, multiservice networks. This concept may actually be realized soon, though ATM will typically occupy only a piece of Layer 2, sharing space with MPLS as the network evolves.
Either way, look for an increased emphasis on “optical intelligence” at the core and “packet intelligence” at the edge. Each carrier will define those boundaries for itself, creating ample niches for startups and ample confusion for big vendors, who have to craft network solutions for their customers without appearing to contradict their own rhetoric. I guess that makes them the real politicians.
Scott Clavenna is president of PointEast Research and director of research at Light Reading http://www.lightreading.com