Has West Bay Got the Best Way?
Why? Apathy? Maybe. But it's also possible that they were dumb-struck by West Bay's achievement and didn't want to give the startup's developments any further oxygen of publicity.
It looks as though West Bay might be delivering on its claims of having a revolutionary chip architecture, one that performs complicated tasks with fewer transistors. In doing so, it sidesteps the usual obstacle facing chip makers trying to take another leap forward in integration -- namely that there's only so many million transistors you can pack into a piece of silicon (see Startup Claims Chip Breakthrough).
On the face of it, West Bay appears to have left the big chip vendors in the dust. Its channelized framer chip, the WB1501, can aggregate 16 OC3 or STM1 (155 Mbit/s) channels and aggregate them into an OC48 or STM16 (2.5 Gbit/s).
"The closest competitive solutions have only four ports, so right off the bat we're four times denser," claims West Bay's president Tino Varelas.
Replacing four 4-port chips with a single 16-port chip results in big savings for system vendors. West Bay's WB1501 costs $440 when supplied in volumes of 10,000 or more, only slightly more than competitors' four-port chips, says Varelas. As a result, silicon costs are reduced to "one quarter to one third" of what they would have been.
West Bay's chip also has on-board clock and data recovery (CDR) circuits, which allow it to interface directly to the optics coming onto the board. This eliminates the need for a PHY (physical layer device) chip and pushes the cost savings up a notch higher.
And that's just for starters. The WB1501 consumes a correspondingly low amount of power, 5 watts. And, of course, replacing four chips with one also enables system vendors to improve the density of their equipment, which is music to the ears of many service providers.
Density aside, the chip boasts other features not found elsewhere in merchant silicon, claims Varelas. For a start, it can groom an entire OC48, something that four-port framers cannot easily do. That means it can reorganise data streams at the STS1/VC3 (51.8 Mbit/s) level -- a key feature for equipment that handles more than one data protocol.
"Let's say you're making a multiservice box," explains Varelas. "What you need to be able to do is aggregate the telecom traffic and the datacom traffic, so it facilitates processing in the next part of the system."
A group of four 4-port framers cannot easily do this, even if they support time-slot interchange. Individual chips could reorganise their own payload, but reorganising an entire OC48 would require the ability to exchange data between chips, which complicates things considerably.
Another significant feature of the chip is the fact that it does full overhead processing. The WB1501 does bit error rate checking on every path through the chip, once for each STS1.
In spite of doing all of this complicated stuff, West Bay isn't pushing the limits on semiconductor technology, thanks to its novel chip architecture. Its chip is being made in 0.25 micron technology, not the 0.13 micron state-of-the-art technology that's more likely to experience manufacturing hiccups.
West Bay is already shipping samples of its chip to customers (see West Bay Intros Processor). Varelas is keen to point out that the chip is sampling in Europe and Asia as well as the U.S. That's because it offers full support for SDH (synchronous digital hierarchy) -- the non-American version of Sonet (see Sonet (Synchronous Optical NETwork) and SDH (Synchronous Digital Hierarchy)).
"I think it's the only device that fully supports a tandem connection," says Varelas. A tandem connection is an SDH-based concept, devised so that service providers may monitor the performance across a particular portion of a network, for example, within one country in Europe.
Ironically, the main snag with West Bay might be the lead that it's stolen on the competition. System vendors aren't likely to find an alternative supplier of 16-port framer chips any time soon -- and relying on a single startup to supply key components could be considered a risky proposition.
— Pauline Rigby, Senior Editor, Light Reading