Matisse Primes Metro Ethernet Makeovers
The 65-person startup says its new EtherBurst products can scale metro networks beyond 10 Gbit/s with gear that's easier to manage and less costly than anything currently sold by Cisco Systems Inc. (Nasdaq: CSCO), Fujitsu Network Communications Inc. , and Nortel Networks Ltd. . (See Matisse Aims to Make Impression.)
Matisse's distributed switch comprises two products: the SX-1000 Ethernet Service Node and the PX-1000 Photonic Node. They use a technology called optical burst switching to improve how the bandwidth in a fiber ring is utilized by allowing for superfast setup and teardown of wavelengths.
The PX-1000 provides generic optical transport around a metro ring, and the SX-1000, which contains a packet processor and a "burst" transponder, does the intelligent stuff. Up to eight SX-1000s can be connected to each of four PX-1000s for a total of 32 SX-1000s on a metro ring, yielding up to 640 Gbit/s of capacity.
Briefly, the SX-1000 takes the incoming traffic and sorts it according to destination and quality of service required. These packets are then assembled into "bursts" and sent to their destination in the network using whatever wavelengths are available. The optical transponder in each SX-1000 can talk to every transponder on the metro ring via any available wavelength.
That's especially important when some uneven traffic spike occurs in a metro network -- as often happens these days with user-generated videos and IPTV deployments ramping worldwide. When spikes occur, the Matisse gear surrounding the traffic spike will purportedly know to queue all low-priority traffic heading to the affected node until the congestion clears.
WDM networks, which are mostly point-to-point links, are becoming more and more automated, says Michael Howard, principal analyst at Infonetics Research Inc. , who says this direction was "enhanced by ROADM-enabled WDM gear."
ROADMs can make automatic power adjustments, and they allow circuits to be provisioned more quickly with less manual labor. But "Matisse appears to have developed the next stage in WDM ring/mesh networks with nearly automated WDM traffic management of packets, source-destination delivery paths, and wavelengths," Howard says.
ROADMs "still are a circuit-based, static system that requires manual configuration," says Sam Mathan, Matisse's CEO.
Does this mean Matisse's gear is a potential ROADM killer? Not in the near term. Analysts don't see the demand for ROADMs falling off anytime soon. And, besides, Matisse is focused on metro and core Ethernet applications. (See IPTV Delays Could Hurt ROADM Deployment.)
"Even though Ethernet is emerging as an important way to distribute IPTV in the metro network, this is still a very small part of existing metro networking that relies on quite a lot of Sonet/SDH for IP, voice, and other applications," says networking analyst Michael Kennedy of Network Strategy Partners LLC .
So Matisse's gear might really make more sense in cable networks or a telco's new video overlay network, says Scott Clavenna, chief analyst at Heavy Reading. "You need a network that doesn't have legacy Sonet, SDH, or circuit traffic in it. And that network needs to have a lot of Gigabit Ethernet that has to go in a lot of different directions," he says. Cable networks fit that descriptions, and "they are in the process already of building big optical networks with tons of Ethernet."
Telco video overlays might be another place where Matisse shows up. Clavenna says it's conceivable that Matisse might be an alternative to ROADMs in networks where the carrier is installing lots of Ethernet and WDM and using ROADMs to create a wavelength mesh.
The data center is another possible playground for Matisse, Clavenna notes. "Wherever Force10 Networks Inc. is in a network, this could be a good transport mechanism for that traffic."
Matisse's Mathan, the man who sold Amber Networks to Nokia in 2001, says his company's new products will be generally available later this year and haven't weathered any carrier trials yet. So far, the only place Matisse's gear has been put through its paces is at the Lawrence Livermore National Laboratory (LLNL) .
Absent a marquee carrier customer, there are a lot of claims left unchecked with Matisse's announcement. For one, it's impossible to know if its claims of opex and capex savings would hold up in a large network deployment. And, as Harvard University senior technical consultant Scott Bradner points out: "What is to keep others from matching the cost savings in the future?"
Though not yet in commercial networks, the company's wares -- along with cutting-edge optical gear from several other vendors -- will be on display at Optical Expo 2006 tomorrow and Wednesday in Dallas.
— Phil Harvey, News Editor, Light Reading
Interested in learning more on this topic? Then come to our upcoming conference, Optical Expo 2006. This conference and exhibition will be staged in Dallas on September 19 & 20, 2006. Admission is free for attendees meeting our prequalification criteria. For more information, click here.