The good news about optical networking is that it lets carriers deliver huge amounts of bandwidth. The bad news is that it's hard to get money out of those fat pipes.

Tenor Networks Inc. (http://www.tenornetworks.com) is looking to solve that problem with the launch of its TN250G Optical Service Switch. It promises a slew of features that will allow carriers to provision differentiated services and-crucially-charge their customers for every packet and cell.

Still, Tenor isn't the only optical player looking to help carriers turn a dime from next-gen services, and its product will face competition both from terabit router vendors and from startups working on Sonet multi-service provisioning platforms (MSPPs) for metropolitan networks. What's more: The device won't have some important routing functions when it ships in the fourth quarter; Tenor hasn't yet announced meaningful pricing information; and it isn't in beta trials-so it's not clear that Tenor can really deliver on its performance claims.

The TN250G is designed to function as a network linchpin-binding metropolitan area networks (MANs) to long-haul networks. Service providers will be able to install it either in the central office (CO) or point of presence (POP), where it will interconnect metro DWDM and Sonet hardware to long-haul DWDM equipment using a variety of interchangeable, high-speed interface modules. Tenor claims the device will process ATM, frame relay, IP, and TDM traffic, and use MPLS (multi-protocol label switching) to set up virtual tunnels over the core network, allowing service providers to engineer their networks for higher performance.

But the TN250G is a lot more than just a big network switch. What makes it interesting is its touted ability to help carriers quickly provision value-added data services-and make money from them. That's a huge deal. Right now, service providers are stuck between a rock and a hard place-trying to turn a profit either from nickel-and-dime voice minute rates, or by delivering low margin, fat-pipe, plain vanilla data services.

The TN250G is designed to break that model by adding "service awareness" to the optical network. That sets it apart from traditional DWDM or Sonet optical equipment, which operates at the transport layer and is too dumb to understand what sort of traffic or applications it's carrying, much less give carriers the information they need to charge for it. Tenor says its switch solves the problem by building ASICs into every switch port. They perform three functions: recognizing packets and cells, applying quality of service (QoS) to them, and recording (or metering) them. Tenor claims the ASICs can do this even when minimum size 28-byte packets are being sent at maximum OC-192 line rates.

These features (traffic recognition, QoS enforcement, metering) are complemented by a service provisioning software suite that allows carriers to remotely set up and tear down connections, or make modifications to service classes, from a network management console.

Carriers can use the TN250G's smarts to refine the raw capacity afforded by DWDM, offering their customers contracts that define multiple service classes, for which the carriers can charge different tariffs. That could also allow ISPs do away with all-you-can-eat, 'free lunch' Internet access, by allowing them and to charge tiered prices for different levels of Internet QoS.

It all sounds great, but competitors say Tenor may have bitten off more than it can chew. "It's hard to say what Tenor will deliver, because nobody's touched [the switch] yet," says Larry Lang, vice-president, service provider marketing group, at Cisco Systems Inc. (http://www.cisco.com). "On paper, it does everything, but in practice, such all-in-one hybrids are very hard for startups, because they must be industry-leading on every capability or the value proposition collapses."

The TN250G will face competition from two other classes of equipment. First: terabit and multi-gigabit routers from vendors like Cisco, Juniper Networks Inc. (http://www.juniper.net/) and Pluris Inc. (http://www.pluris.com). Second: Sonet MSPPs (seeSonet Goes POP). Each type of equipment is installed in the same place as the Tenor box: the CO, or POP. And, like Tenor, the vendors selling routers and MSPPs are making a big noise about their products' service-provisioning skills. On paper, though, none of them appear to match Tenor for high-speed metering capabilities.

Tenor says its product will complement these other equipment classes-not compete with them. "Siara [a Sonet MSPP startup] is one of our closest competitors, but it isn't a direct competitor," says Sean T. Welch, vice president of marketing and sales at Tenor. He says the TN250G could be installed in a CO or mega-POP to aggregate traffic from multiple Siara MSPPs. Similarly, Welch claims that carriers will install the TN250G alongside terabit routers, such as Juniper's, in order to add metering facilities to their networks.

One thing is certain: Carriers won't be able to install the Tenor box instead of routers-at least initially. That's because in its first release the device won't ship with support for high-end router functions like BGP (border gateway protocol). "We'll add full IP routing later, when it becomes necessary," says Welch. "We recognize that it's a non-trivial task."

In the meantime, the box will rely on Layer 2 MPLS to parse all protocols out over the core network. Tenor says it decided to go with the MPLS approach eighteen months ago, when it started work on the product. It's a decision that could pay huge dividends for the company, given that MPLS looks almost certain to become the lingua franca of optical networking interoperability standards now being worked on by three different organizations (seeCrunch Time for Signaling Standard )

"MPLS is important for traffic engineering in both IP networks and ATM networks that support IP. We will implement it," comments Fred Harris, director of network planning and design for Sprint Corp. (http://www.sprint.com).

Tenor says the TN250G is a non-blocking switch. In maximum configuration it supports up to 576 DS3 (45-Mbit/s) or OC3c (155-Mbit/s) ports; 192 OC-12c (622-Mbit/s) ports; 48 OC-48c (2.488-Gbit/s) ports; or 12 OC-192c (9.953-Gbit/s) ports. Those ports are connected internally over a 240-Gbit/s switch matrix (full spec sheets are at http://www.tenornetworks.com/TN.htm).

The product will go into beta tests in the second quarter of the year, with general availability in Q4, Tenor says, so there's yet no way to know whether it will perform as advertised. Indeed, Tenor is still in the process of forging its ASICs-the hairiest part of the product development cycle.

And Tenor is being miserly with pricing info. Pricing for the chassis alone starts at $90,000. Interface modules will be sold separately, and the vendor hasn't said how much they will cost.

So what does the future hold for Tenor as a company? Its money-making pitch could easily make for a red hot IPO-if it doesn't get bought up first. It's interesting to note that one of its outside directors is Dan Smith, president and CEO of Sycamore Networks Inc. (http://www.sycamorenet.com). That circumstance, plus the clear-cut synergies between the TN250G and Sycamore's core optical switches-and Sycamore's overflowing post-IPO coffers-has Light Reading wondering whether Tenor won't be quickly acquired by Sycamore.

-Stephen Saunders, US Editor, Light Reading, http://www.lightreading.com