BY MEGHAN FULLER
Today's next-generation SONET boxes combine the functionality of add/drop multiplexers, necessary for ring interconnection, with the functionality of digital crossconnects, required for switching and grooming. While that may sound like "unsexy stuff," says Marian Stasney, senior analyst at the Yankee Group (Boston), "where it becomes interesting is the fact that we are able to pack all of that stuff into smaller and smaller form factors that are more efficient and more manageable." Improvements in silicon have enabled next-generation SONET devices to achieve superior functionality compared to their early-generation cousins.
A combination of technological improvements and the compelling economics of next-generation boxes have given SONET renewed life. According to Stasney, next-generation SONET multiplexers (or "switches," as she calls them) enable capital-expenditure savings in the 30-70% range.
"The reality is, in today's economic environment, what carriers are focusing on is profit," asserts Dave Lively, senior manager of optical strategy at Cisco Systems Inc. (San Jose, CA). "To stay profitable, they have to be able to smoothly move from their existing legacy infrastructure to more of a next-generation infrastructure. You can't just do a wholesale swap out of one technology for another."
The Yankee Group predicts that worldwide revenues for next-generation SONET boxes will experience a 34% CAGR to reach $11.7 billion by 2006, up from $2.7 billion this year. Though such boxes have been carrying limited live traffic since the late 1990s, they are expected to see wide-scale deployment beginning this quarter.
Lively compares early generations of SONET boxes to an army tank; they were solid but rigid and inflexible. "Traditional SONET multiplexers had multiple platforms, each optimized at different rates," he explains. "You'd have an OC-3 platform, an OC-12, an OC-48, and an OC-192, and they'd all be optimized for their different rates. They weren't flexible in terms of being able to drop a variety of rates or services at any given location."
Advancements in silicon have enabled today's next-generation SONET boxes to far exceed earlier incarnations. "Traditionally, you had one chip doing 10-Gbit framers," explains Jack Basi, vice president of marketing at Infineon Technologies Inc. (San Jose). "We have announced a device that is 4 x 10 Gbits on a single device. Obviously, if you can incorporate that kind of high-density silicon, you can reduce your footprint, you can reduce your power consumption, and therefore reduce your cost."
"The reductions in cost and space are just overwhelming," agrees Stasney. "A popular configuration of the traditional DCS [digital-crossconnect system] took up 27 bays for 1,024 STS-1s. [It cost] $5 million to set the first bay on the ground and probably up to $15 [million] to complete the whole installation." Stasney adds that 150 man-hours were required per bay. "Those numbers really knocked me off my feet-when you can do the same thing [now] with density at the bare minimum four times greater than that for less than half-a-million dollars in less than half a bay," she concludes.
According to Basi, another trend he is seeing at the silicon level is a chip capable of accommodating both SONET and SDH, enabling operators to configure their boxes either way.
Silicon advancements have also enabled integration, which has led to increased functionality. For example, today's next-generation SONET boxes more closely resemble small crossconnects than do traditional add/drop multiplexers (ADMs), because they enable "any port to any port" connectivity-something their early predecessors did not. From 1988 to 1998, SONET boxes relied on a time-slot assignment (TSA) switch fabric, which was far more limiting.
"It didn't allow you to do hair-pinning, for example," explains Greg Wortman, vice president of marketing at Coriolis Networks (Boxboro, MA), "which is where you would take a circuit inside a node and loop it back out of the same node." But by the late 1990s, SONET boxes began to sport a time-slot interchange (TSI) switch fabric, which enables operators to groom traffic between rings or groom traffic from a collector ring to a higher-rate distribution ring-all in a nonblocking manner.
Perhaps the most important characteristic of a next-generation SONET box, however, is the ability to officially carry data services over a SONET infrastructure-though the best method of delivering those services remains a point of contention among industry players. Today, vendors employ various methods for carrying or mapping data streams, particularly Ethernet, over a SONET infrastructure, including the generic framing protocol (GFP), virtual concatenation, and the ITU's X.86. GFP encapsulates just about any data protocol into a SONET frame in a standard way, enabling traffic to orginate on one vendor's equipment but end up at another vendor's equipment. Virtual concatenation enables vendors to use arbitrary concatenations over SONET without affecting any of the installed legacy gear in between, while X.86 defines a standard for mapping Ethernet into STS-1 pipes.
A common standards-based transport protocol would facilitate integration, contends Infineon's Basi. "Currently, most of the rings are single-vendor, so the advantage of this would be to create a multiple-vendor single ring. It would allow a great deal of flexibility for carriers," he says.
Just which protocol or protocols will become dominant is difficult to predict, admits Pat Hatton, senior analyst in optical transport and networking at RHK Inc. (San Francisco). However, she adds, "If a large number of carriers start favoring one method or vendor, that method will quickly become the de facto standard. A developmental platform will fail if it doesn't get market traction from large carriers, even if it is a better solution." On the other hand, she says, if a single large carrier makes a commitment to a vendor to deploy a product that supports, say X.86 or GFP, then the chances of that technology continuing in the market are much greater.
Some SONET boxes have so much built-in functionality, they have earned the nickname "God Box." They provide transport, including DWDM and ADM capability as well as ATM switching or IP routing-and sometimes both. But is there such a thing as too much functionality?
Hatton contends that some carriers may view the God Box (or "omni box," as RHK calls them) as a single point of failure. "The complexity and multiple functions will be a deterrent for the incumbent LECs [local-exchange carriers] who would have to make enormous changes to current methods of operations support systems and processes," she asserts. "In the larger and more complex legacy networks, it may be more economical to carry each traffic type on a separate overlay network."
That said, Hatton also admits that the omni box is likely to appeal to newer carriers with greenfield-type networks, where it is easier to realize the economic advantages of replacing many products with a single product.
But to those startups and emerging vendors considering this approach, the Yankee Group's Stasney has a few words of warning. "My recommendation would be to go that way with a lot of carrier input under [your] belt, because that's not what I'm hearing from the carriers." They want their functionality to be more modular, she says, because it affords them greater flexibility. It also provides them a wider choice of vendors, so they are not caught in a single-source or proprietary solution.
Much has been made lately of the so-called battle in the metro between next-generation SONET, Ethernet, and DWDM. While some, particularly those who champion Ethernet or DWDM in the metro, argue that next-generation SONET is a short-term fix to the long-term problem of data transport, the SONET vendors believe that SONET is here to stay-at least for the foreseeable future.
It all comes back to cost and manageability, says Cisco's Lively. "Why is SONET still here? And why will it continue to be here for easily the next decade?" he asks. "Because that's what's out there and installed. It is far cheaper for carriers who have an existing legacy using SONET to keep deploying SONET services for next-generation applications. They bridge the gap between a legacy infrastructure and what we know the full-service network is going to be moving forward."
Stasney contends that SONET and DWDM can coexist because the latter is merely a transport technology. "We can add DWDM either to the back or to the side of next-generation SONET boxes in order to transport packets more efficiently," she says. "The technologies are definitely complementary."
And while Ethernet may prove to be an enabling technology going forward, its current price points limit the level to which it can compete with SONET. "While Gigabit Ethernet is very exciting and new and getting a lot of press, as far as the actual amount of revenues it's bringing to the operators-especially the ILECs-it still remains a minority in relation to the revenues from traditional SONET," asserts Tom Fuerts, director of the terrestrial networking division's global solutions at Alcatel (Paris).
Also emerging as a viable option for the metro is resilient-packet-ring (RPR) technology, and while some in the industry believe it will directly compete with SONET, others believe they can coexist. "RPR seems to be a unifying standard that will allow the extension of Ethernet but in a carrier-grade way," says Basi. "It allows SONET to be redefined for the metro space, while preserving the underpinnings of SONET and unifying Ethernet."
Coriolis's Wortman, on the other hand, believes that SONET remains the best option. "It's kind of like comparing a tow truck to a car-hauler-a transport that carries lots of vehicles on it versus a tow truck that carries just one vehicle," he contends. "RPR only handles packet traffic. It doesn't handle TDM traffic."
According to RHK's Hatton, there is no one-size-fits-all solution for the metro. "Where the technology has the features, functionalities, and economics that fit with a carrier's strategy and business plans, that technology will be deployed by that carrier," she asserts.
Download SONET multiplexers PDF file (file is 28k)