TeraBurst enters optical-switch fray with millimeter wave technology

BY MEGHAN FULLER

Crucial to the realization of the all-optical network, the optical switch transforms raw capacity into manageable and revenue-generating bandwidth. The problem with existing switches, say representatives from TeraBurst Networks Inc. (Sunnyvale, CA), is that an optical-electrical-optical (OEO) switch offers superior signal quality, while an all-optical (OOO) switch enables line-rate switching and low latency, but neither does both-which is what today's carriers require.

"We've talked to a lot of our customers and looked at a lot of market research, and these are the key functionalities that they want out of their optical switches: scalability, speed, flexibility, reliability, quality of service, and provisioning," explains Rich Lee, strategic marketing manager at TeraBurst.

Several switch vendors are currently developing hybrid solutions designed to combine the intelligence of digital technology with the speed of photonics-without much success, claims Dr. Shantanu Mitra, TeraBurst's di rector of marketing. "The push is on toward getting a hybrid solution together," he says, "but the labors of the OEO and the OOO are still distinct and just jammed together. Our approach gives you the functionality that you're looking for from the OEO and the OOO in the same switch fabric."

The OMS 2100 and OMS 2200 switch whole wavelengths-in the analog domain. The signal travels through the OMO switch and into the line cards, where it undergoes an optical-to-electrical (OE) conversion, which turns the signal into a high-bandwidth, transparent, analog millimeter wave. "The key," says Mitra, "is that it's sent through the switch fabric really as a linear form, so it's much like the photonic in that it's a space switch." Because there is an OE conversion at the line card, he explains, the switch is also capable of 3R regeneration (retransmit, reshape, and retime), SONET/SDH performance monitoring, and forward error correction. The switch fabric is protocol independent up to 40 Gbits/ sec-but it's not the only element that can handle 40-Gbits, says Mitra.

"In terms of an upgrade path, you have a very interesting approach to upgrading to 40 Gbits or OC-768. When the line card optics are commercially available [at OC-768 speeds], it's just a matter of replacing the appropriate line cards, keeping the same box, the same backplane, the same switch fabric. It is not a forklift upgrade," he adds.

The OMS 2200 switches wavelengths at line rates from OC-3 to OC-192 and scales from 640 Gbits/sec up to 64 channels in a half-rack footprint.

Targeted at regional/metro markets, carrier/data hotels, and interconnectivity between enterprises, the OMS 2100 is housed in a one-third-rack footprint, handles 200 Gbits/sec today, and is scalable at OC-768 speeds to 800 Gbits/sec.

Company representatives are quick to point out that neither OMS switch is merely a dumb crossconnect; TeraBurst has developed software targeted at both the element and the network level to enable end-to-end provisioning, flexible scheduling, and reservations per user requirements.

Included in the OMS SoftSuite are element controllers, which reside on the switch itself and control each individual network element. An element-management system (EMS) then controls several of those network elements through an EMS server that communicates with the various elements through 10/100 Ethernet lines.

TeraBurst has also developed a Tier 2 network-management system (NMS) that does the wavelength routing and protection within a subnetwork of the company's products. "We can do end-to-end provisioning within a cloud of TeraBurst switches," says Mitra. "We do that with easy-to-use GUIs [graphical user interfaces] through point-and-click provisioning. All of that then uploads into a Tier 1 NMS that is typically resident on the carrier's system. And we have CORBA and Java interfaces to allow that to happen," he adds.

On paper, TeraBurst's OMO technology seems promising-so what's the catch? The catch, says analyst Grier Hanson of Current Analysis, is that this application of the technology is completely new. "It hasn't been proven, it's not being used by anyone else, and it's just getting set to enter into trials right now," he says.

Proving itself to carriers and service providers may not be an easy task, says Mark Lutkowitz, vice president of optical-networking research at Cir Inc. (Char lottesville, VA), because service providers may have "a bias against analog technology. There was a company a while back called Silk Road that used analog technology and nobody could figure out the patents," he asserts. "I'm not saying that [TeraBurst] isn't for real, but there is a knee-jerk reaction when service providers hear analog technology." He also cites the less-than-robust field of optical frequency division multiplexing (FDM), which also employs analog technology.

TeraBurst, anticipating exactly these concerns, commissioned a third party to conduct an availability and uptime analysis of their system to calculate the meantime between failures (MTBF) of the individual components. "The analysis was done using standard Telecordia methodology," explains Mitra, "using markup models to come up with the number of failures in 10⁹ hours. The result of that analysis shows a five-nines-plus availability for the entire system based on MTBF numbers."

For its part, the company isn't worried that the obstacles are too difficult to overcome and is moving full-steam ahead. Since its inception in January 2000, TeraBurst has raised $51.4 million in venture financing and has already received four letters of intent from potential customers. TeraBurst has also entered into a pair of strategic alliances, including one with Atoga Systems (Fremont, CA), for sales and marketing.