Interesting article. fiberopticsonline.com{581BC931-99E2-11D3-9A6C-00A0C9C83AFB}&Bucket=HomeFeaturedArticles
Next Generation Networks Event Chronicles Optical Advances
11/15/99 Optical networking emerges as this year's hot topic at the Next Generation Networks conference. Presenters chronicle advancements in the field and assess the impact of the next generation of activity.
By: Erik Kreifeldt
In an introductory presentation for this year's Next Generation Networks Conference (Washington, DC, Nov. 1-5), conference chair John McQuillan identifies optical networking technology as one of the top three technology forces shaping next generation network infrastructure, along with integrated circuit and packet routing advancements. During an optical networking symposium on the last day of the conference, presenters from optical metro transport start-ups tell the sexiest stories.
Revolutionary breakthrough
"DWDM is really a breakthrough technology in just the same way as we see in chips?only faster," says McQuillan, a venture capitalist and Harvard-educated networking guru. Now optical technology is emerging to make "virtual trunks" out of wavelengths, he postulates.
The old networking model was a static, off-line, trunk circuit configuration that featured dynamic, real-time packet routing, McQuillan posits. The new model consists of a dynamic, real-time configuration, he explains, with static, off-line, routing. "This is the first really big change in networking I've seen in quite a while," he proclaims.
Optical networking capacity is escalating in both channel count and bit rate, McQuillan notes. The next big thing after DWDM may add another dimension to this scale?optical multiplexing. He cites three proposed flavors: Coherence division multiplexing, code division multiplexing, and frequency division multiplexing.
"My forecast is still for exponential Internet growth [in capacity]," McQuillan says, noting that the top technology companies are devoted to the cause, along with start-up companies funded by billions in venture capital. He calculates that optical fiber communications capacity has increased at a rate of 100% per year, that demand for data is increasing at 300% per year, and that both venture capital funding and industry acquisition value totals are increasing at a rate of 100% per year. The next decade will be the golden age of network expansion, application development, and mass deployment, McQuillan concludes.
Intriguing optical stories
As part of an optical networking symposium on the last day of the conference, presenters toss about parameters such as optical versus electronic switching, transport bandwidth granularity, and trade-offs between bandwidth and distance.
Executives from Ciena Corp. and the company's new customer, Williams, anchor presentations on state-of-the-art optical networks, while Qtera and Corvis founders speak to all-optical, ultra-long distance network architectures. Ciena's Steve Alexander, VP and chief technical officer, acknowledges that the physics for 16 Tb/s transmission is available, but customers are not demanding it yet. In the meantime, adding line cards to existing systems comprises a brisk business, he says.
Metro optical transport
Jonathan Reeves, president and CEO Sirocco, chronicles new solutions to metro network problems. He breaks the field down into four categories: First and second generation optical metro equipment, and SONET-centric and packet- or cell-centric next-generation add/drop mulitplexers (ADMs).
First generation optical metro transport features one wavelength per slot, simple ring topology, no packet or circuit switching backplane, and only accommodate packet services with gigabit Ethernet interfaces on its wavelength channels. Second generation systems sport more wavelengths and more complex topologies, built-in SONET protection switching, improved network management, dynamic optical power control, simple SONET multiplexing per slot, and no packet or circuit switching backplane.
Packet- and cell-based next-generation ADMs rely mostly on an ATM fabric, Reeves says, with virtual path rings the most common approach. Cisco Systems' Dynamic Transport System is an example of a packet approach. The systems are not well suited to circuit-based optical cores or delivering raw bandwidth, Reeves asserts, and optical infrastructure is limited.
SONET-based next-generation ADMs, such as the Cerent 454, are more data-capable than traditional ADMs, Reeves notes. The network elements feature an integrated packet and circuit switching backplane and, like packet- and cell-based ADMs, have little optical infrastructure.
Reeves proposes an "optical edge switch" that performs SONET aggregation and grooming, wavelength virtual private networks, transparent bandwidth provisioning, ATM access concentration, IP VPN and Internet access services, and ring or mesh architecture.
Optical submultiplexing
All-optical approaches tend to devote one wavelength per service, which makes economic sense when delivering services that require optical-class bandwidths of hundreds of megabits per second, presenters say. If a metro transport system performs electrical submultiplexing of wavelengths to accommodate several narrower-band services per optical wavelength, the question becomes which framing protocol to use, notes Reeves. ATM or SONET or IP make sense for different applications, he says, adding that the most flexible system should establish a framing protocol on a case-by-case basis.
Chromatis's chairman and CEO Rafi Gidron concurs with the submultiplexing approach. Although multiplexing with various framing protocols and interfaces is complex, the technique pays off with efficient bandwidth utilization, he says. Gidron's company differentiates its metro DWDM system with a submultiplexing strategy that cost-effectively accommodates optical network nodes that do not demand optical-class bandwidth.
New metro carriers experience "outpost growth," postulates Raj Shanmugaraj, president and CEO of Astral Point. This begs a mesh network topology, he says, as opposed to the optical rings associated with metro networks today. Rather than build out a well-planned network of central offices placed 18,000 feet apart over time, these carriers place facilities wherever they can find the real estate?in the basement of a multi-tenant building or hotel closet, for instance. Then the carrier sells services to business within striking distance of the real estate. A metropolitan communications system must scale to meet these needs on a just-in-time basis, he says, adding that the growth pattern is not conducive to ring architectures.
"I don't think rings will go away any time soon," Shanmugaraj says. "A flexible topology is what's needed." He proposes a mesh architecture of integrated electrical/optical network elements that integrate functions of digital cross-connects, ATM switches, local area network and packet switches, and optical wavelengths, handling signals from DS-1 to OC-192.
Ring-based optical transport
Optical Networks president and CEO Hugh Martin defines the metro core as networking OC-3 through 150 Gb/s worth of traffic, and the metro access network as handling T-1 through 10 Gb/s traffic. In the metro core, Martin makes the case for optical transport, maintaining that optical-to-electronic techniques lend themselves better to access networks, with their T-1-class interfaces.
Although focused on the metro core, Martin reports that Optical Networks has entertained inquiries from 28 enterprise users, who are interested in networking gigabit chunks of data, which he labels "OC-24."
Martin cites a field trial application that operates six channels of OC-48, four channels of gigabit Ethernet LAN extensions, and two wavelengths of Fibre Channel. The carrier also wants the traffic protected, with dynamic add/drop capability. An ATM core could provide the performance, he acknowledges, but maintains that allocating 12 channels of a metro DWDM system with optical protection and add/drop proves to be more cost-effective.
One cost analysis that Optical Networks performed for a regional Bell carrier concluded that a metro DWDM system prove to be more cost-effective than OC-48 and OC-192 ADMs to provide OC-12 data rates on a typical five-node ring.
Ring or mesh, optical or electronic, optical networking commands big interest among networking industry watchers. "Optical networking is the epitome of creativity and innovation in our field," McQuillan surmises in a conference-closing remark following the optical networking symposium.
|
