Matisse Networks exited stealth mode in September, introducing what may be the first commercial application of optical burst switching technology to telecom networks. Using a combination of optical networking and Ethernet switching, Matisse’s Etherburst platform rapidly routes packets to different wavelengths, setting up and tearing down optical links in nanoseconds, thereby combining the scalability of optical networking with the efficiency of packet-switching. Just before Thanksgiving, Matisse’s chief executive officer Sam Mathan told Telephony’s Ed Gubbins where Matisse is headed next.

On customer traction: We have finished our first customer trial and are now in the planning stages for a second trial. Our first beta customer was a campus. The customer trial subsequent to that was with an Internet exchange peering point. The second customer trial will be with a [competitive local exchange carrier] that also is a leading provider of business Ethernet services. We expect our product to be generally available by the end of this quarter.

On sales and marketing efforts: Matisse’s macro plan for sales and distribution is to partner to achieve scalability, partner with large equipment providers that have a need to serve customers with total solutions that are Ethernet-centric. If you look at the [dense wavelength-division multiplexing] transport market, there’s very little organic production left that’s Ethernet-centric. The [reconfigurable optical add/drop multiplexers] of Fujitsu and Tellabs are the only new generation of ROADM products. We hope to fill that gap in distribution for larger providers wherever they are in the world. Our initial focus is to serve the U.S. market, establish ourselves with reference accounts and move on in 2007 to Europe and Asia.

On sales force: Like other startups, we’re a highly engineering-centric company. We have 65 employees, of which 55 are in engineering. About 8 are in marketing and sales. Of course, everybody in the executive ranks is in sales and marketing, customer-facing. We’ve been very careful in building our sales organization over the last year or so. [On November 20, 2006] we hired a vice president of U.S. and European sales: Doug Stewart, the former VP of sales for TiMetra. He was also VP of North American sales for Amber Networks, where he worked for me.

On enterprise versus carrier customers: The Etherburst product has a significant market opportunity, both with service providers and enterprises. In the early stages, I expect enterprise or private metro networks will probably be the predominant part of our revenue stream--that could include enterprises, ISPs that have dark fiber, government agencies, hospitals. Because of their longer sales cycles, carriers will be more dominant as we continue to penetrate that market. As [carriers] displace the 400,000 Sonet aggregation rings over time with Ethernet aggregation that can carry voice packets or TDM over pseudowires, we expect those all to be solved with products like Etherburst. We’re seeing traction wherever people are providing Ethernet aggregation--CLECs or local exchange providers for IPTV or businesses moving more toward packetized voice, high-quality video for surveillance and interconnection of campuses in multi-location networks.

On Bells: I’ve been working with [regional Bell operating companies] since 1995 and have a lot of relationships. We’ve been able to have a continual dialogue. We’re in early sales cycles with all of them. RBOCs, local exchange providers and PTTs are all moving to Ethernet as their interface of choice. AT&T’s Lightspeed project is all based on Ethernet aggregation. SBC is already offering a series of Gigaman high-performance Ethernet services. As they expand the footprint of these services, they’ll continue to move toward Ethernet.

On secret sauce: The core intellectual property at Matisse is multi-fold. The most important objective for Matisse was to get an optical systems product to market with the lowest possible investment [required from customers] and the ability to match the cost of current transponders, which are tunable for reconfiguring purposes. We invented our Tango 10-Gb/s optical burst transponder and implemented it using commercially available silicon and laser technology. That laser technology is from JDSU Agility--the same laser used in some of the ROADMs. However, our engineers invested in an algorithm capable of allowing the laser to switch wavelengths in the C band in nanoseconds and recover data and clock in nanoseconds in bursts of light. We have the same component cost base as everybody else’s transponder. And we need only one transponder from each location to communicate with other locations. Our other intellectual property is a mechanism to use the color of light as an addressing scheme when you’re in the photonic layer. That allows us to burst colors of light in a single optical hop to any other transponder in a ring. We’re setting up virtual circuits and tearing them down in nanoseconds. We have filed for 7 patents and we have already been awarded 2 of those.

On whether Matisse’s gear can handle, for example, HDTV traffic fast enough to allow a good user experience: The answer is yes. Our platform has two essential components: a photonic layer and an intelligent Ethernet switching function which operates like any other Ethernet switch. With [AT&T’s Project Lightspeed], you have Alcatel’s 7450 Ethernet switch connected to Alcatel’s 7750 router. The way the connections are done in the current architecture is using the 7450 on point-to-point DWDM connections. We provide packet-switching infrastructure that uses the efficiency of optical bursting, the statistical effectiveness of the Ethernet domain similar to what the 7450 does.

On competing with metro ROADMs: The application that our optical Ethernet switch is targeted for and provides huge opex and capex advantages for is the aggregation of Ethernet traffic. [AT&T] is deploying Fujitsu [ROADMs] to connect Alcatel’s 7450s and 7750s. All service providers, whether telcos or cable providers, are moving toward converged networks where they can carry IPTV traffic. In today’s packet-oriented service and transport networks, it’s extremely costly and inefficient to try to have a converged optical layer. Etherburst allows the building of an optical layer that matches the flexibility of Ethernet to serve those applications. We will compete with ROADMs in those applications.

On funding: We raised $21 million in 2003 as a Series A. In July we closed a $13.5-million first round. We’re working to complete that round with $17 million. It can happen anytime. We hope to have it in done in three to six months. We have adequate cash.

Want more?

Check out what your colleagues are reading on Telephony's Most Popular.