Briefing.com's Photonics Revolution VI: Opticon2000
(for chapters I-V, see post 28976 above)
[BRIEFING.COM - Gregory A. Jones] Last week's first ever Opticon convention presented by BCR and Light Reading gave us a chance to pause and reflect a bit on the state of optical networking. We all know that this sector is booming -- both in terms of the underlying business and the market valuations. But just where are we in this revolution and where are we headed? In Part VI of our series, we will take a break from our navigation of specific optical sectors and focus instead on a broader view.
Note: Briefing.com subscribers can access parts I through V of the series via the links below. If you are not a subscriber, please take a free trial at www.briefing.com so you can access our Archives.
Photonic Revolution I: The Fiber Landscape
Photonic Revolution II: DWDM
Photonic Revolution III: Optical Switches
Photonic Revolution IV: Transmission and Components
Photonic Revolution V: Terabit Routers
In looking at the big optical picture, it is once again necessary (as it was in Part I) to break the optical sector down into its parts. The reality is that the optical revolution is progressing at vastly different rates depending on what part of the network you are talking about. At the core, photonics are fast winning the day. But at the edge, fiber is still something you get in a bowl of cereal, not something you see in your local network.
Core
At the core of the network, delivering voice and data traffic via fiber has been a reality for years. The transition underway is not a change in the medium, but a change in the capacity of that medium. As we have discussed in Parts II-V of the series, the bandwidth capacity in fiber at the core is soaring due both to improvements in the fiber itself, the advent of DWDM, and better component technology.
Not only is raw bandwidth exploding, but the introduction of optical switches and terabit routers is doing a respectable (though not complete) job of routing signals. Yet even at the core, the all-optical network is not a reality. All-optical switches are now in field tests and are expected to be commercially available late this year or in 2001. But even optical switches will not make the core of the network all-optical. Optical switches have no routing capabilities, and thus some equipment will eventually be performing OEO (optical-to-electrical-optical) conversions, be it terabit routers for IP traffic, ATM switches for ATM traffic, or digital cross-connects for voice traffic.
The network core will only become all-optical when we have the ability to look inside of a light signal and read the bits, similar to what can now be done with an electron. This technology does not yet exist, and might not for several more years.
Metro
The MAN (metropolitan area network) is in rapid transition to something that more closely resembles the core. Fiber rings are proliferating in the MAN, which is a first and necessary step. DWDM, however, is just now arriving to the MAN, and it's not clear what role it will play. Current DWDM technology has generally been too expensive to deploy in the MAN.
Unlike the core, the massive amounts of bandwidth offered by DWDM on a point-to-point basis can be squandered in the MAN, where demand for bandwidth might be huge in one part of the fiber loop, making DWDM cost-effective, but minimal in another part, making DWDM a huge waste of money. Unfortunately, knowing where demand will pop up is hardly a science, and thus DWDM deployment in the MAN has been slow. There are solutions being offered to this problem (more on this in a future MAN installment), but suffice it to say that the bandwidth explosion at the core has not yet been replicated in the MAN.
When DWDM does invade the MAN, the need for switching and routing equipment that can handle the load will escalate, just as it has in the core. At this point, metro demand takes the form of digital cross-connects for voice and gigabit IP routers and ATM switches for data. As DWDM expands metro bandwidth, optical switches and terabit routers will make their way from the core to the MAN.
Though the metro is lagging the core, its development will in general terms be similar. Fiber is proliferating, DWDM is coming, and high bandwidth optical switches and terabit routers will ultimately be needed to handle the traffic. But we're probably 1-2 years away from this market becoming significant.
Access
There are really two features to the MAN -- the backbone, which is the fiber and equipment that delivers light signals through the MAN. And then there is the access equipment that delivers the MAN's bandwidth to its customers, in the form of services such as voice, data, VPNs (virtual private networks), etc.
The access market is expected to greatly exceed the size of the MAN transmission market, which helps to explain Cisco's (CSCO) $7 bln purchase of Cerent last year. Cerent's 15454 box is an access product that aggregates voice, data, and video traffic in its many protocols (IP, ATM, DS1, DS3) for transport over a SONET infrastructure (SONET is sychronous optical network, the legacy protocol that was designed for transporting voice traffic over fiber). Despite endless SONET obituaries, this protocol still dominates in the nation's fiber infrastructure. Hence the tremendous success of Cerent's product, which Cisco believes will generate over $1 bln in revenue over the next year.
Cerent is certainly not alone in this market, though its focus on a product that works well within the current SONET infrastructure has made it an early leader. There are many private companies also vying for this business, and employing a number of different technologies to win. Some companies see Gigabit Ethernet switches ultimately providing carrier class services (thus replacing or supplementing SONET), others look to passive optical networking, and more. We will look at these options in more detail later in the series.
The key message from Opticon concerning the access space was that those companies such as Cerent, that leverage the existing telecom infrastructure will fare best in the short run. But the low cost of Gigabit Ethernet and soon-to-come 10 Gigabit Ethernet (products coming next year) will challenge SONET solutions in the long run. New service providers such as privately held Cogent that have no legacy infrastructure are opting for the Ethernet strategy and, in Cogent's case, believe that an Ethernet solution will make possible 100 mbps service at the staggeringly low price of $1000/month. This type of pricing offers the key to disrupting the entrenched technology. Service providers looking to protect legacy SONET equipment risk being unseated by newcomers who leverage new and more efficient technology.
Edge
The edge is that place where a user actually tries to tap into the telecom network, whether it's by clicking through to Briefing.com, picking up the phone to make a call, or a corporation backing up several terabits of data with StorageNetworks (STOR) services. The Edge is where many businesses and consumers have suffered and continue to suffer from bandwidth starvation. The edge was the black hole at Opticon.
We say that it was the black hole because it was hardly discussed, and what few discussions that did arise tended to be of the war-story nature about how tough it still is to get bandwidth in the real world even as the likes of Corvis (CORV) achieve stratospheric valuations.
The harsh reality is that all of the wonderful technology being discussed at Opticon assumes something that generally isn't available -- fiber to the edge. This means fiber into the office building and residential dwelling. It was noted that roughly 7% of commercial buildings currently have fiber, and an optimist suggested that since these buildings were likely among the largest in terms of square footage, perhaps as much as 25% of businesses might have access to fiber. Judging from the anecdotes, this figure seems a bit optimistic.
As a result, we continue to have a huge mismatch between the bandwidth that can be delivered over the core of the network and that which can be accessed at the edge. This might well become the key issue for the optical networking sector in the coming few years.
Supplying The Edge
It's arguably the case that the DSL and cable modem connections that were supposed to be the answer to the bandwidth dilemma may not be enough. The killer app of broadband that Robert V. Green has been noting for the past year -- on-demand video -- is not something that will flourish at the current speeds of DSL and cable connections. And this killer app is the one that virtually ensures that despite the massive increase in bandwidth supply, bandwidth demand will continue to be insatiable for years to come.
But how are we going to deliver 100 meg connections to homes and up to 1-10 gig connections to businesses? Fiber is the best answer, but it's also the most expensive. Fixed wireless and free space optical (the former uses radio frequency technology, the latter lasers) will be cheaper, but reliability, bandwidth, and coverage area all remain key issues.
Economists have an old theory called Say's Law, which claims that supply creates its own demand. This is likely to be true of the optical networking business, but only when the supply of huge bandwidth reaches the end-user, be it at home or work. Though Opticon had great news about the core, metro, and access markets, the edge was rarely discussed, and looms as the key question. It will happen, but when is uncertain.
Investors in this sector need to be mindful of this issue, because one risk is that optical equipment companies produce a bandwidth boom at the core only to find that access to that bandwidth at the edge is not yet available. We will be keeping an eye on developments at the edge in the months ahead.
Greg Jones - gjones@briefing.com |
