Packet Attack Changes Chip Designs
The pull goes in two directions. Chip advancements are making some new telecom ideas possible. At the same time, the changing face of the telecom industry -- the migration to packet-based networks and the recent obsession with broadband access -- is prodding new chip designs.
Either way, the end result is a new wave of integration. That's particularly true in Sonet (Synchronous Optical NETwork) and SDH (Synchronous Digital Hierarchy) metro networks, where boxes such as the multiservice provisioning platform (MSPP) or add-drop multiplexer (ADM) are being squashed onto single chips. Startups such as Parama Networks Inc. (acquired by Bay Microsystems Inc.), Cortina Systems Inc., and Crimson Microsystems Inc. are working on compact metro chips that combine multiple functions and protocols (see Parama Sells for a Song, Cortina Releases Arsenal, and Micro MSPPs Are Big).
That moves them into territory traditionally held by Applied Micro Circuits Corp. (AMCC) (Nasdaq: AMCC), PMC-Sierra Inc. (Nasdaq: PMCS), and Vitesse Semiconductor Corp. (Nasdaq: VTSS).
Chip makers targeting upgrades to Sonet/SDH networks face some challenges not necessarily found in the Ethernet realm. Any chip aiming for a Sonet socket "is going to have to fit into these platforms that are already deployed and have been out there for a while," says Gerry Jankauskas, chief technical officer of Bay Microsystems.
That kind of thinking should be designed into chips such as integrated framer/mappers, according to Jankauskas. "If [equipment vendors] can maintain a hardware design through several iterations of system releases, that's probably the most critical piece," he says.
Chip makers are also getting more involved with the end products, often by providing boards or even helping with systems design. "We're re-forming our organizations to provide consulting services to assist equipment providers," Jankauskas says.
With chip integration on the march, systems themselves are shrinking, upsetting the familiar split between chassis-based systems and fixed-format "pizza box" products. "What we're seeing are a lot of systems that are in the middle," chassis boxes that are squashed to small sizes, says Simon Stanley, the Heavy Reading analyst moderating the sessions here.
Some of these newer designs point to an acceptance of the Advanced Telecom Computing Architecture (AdvancedTCA), the set of emerging standards for telecom hardware. Big names like and have expressed support for ATCA. But in general, it's the smaller equipment providers that will embrace ATCA first. "The major TEMs are going to hold onto their proprietary [formats] as long as possible," says Alexander Smith, platform solution architect for . (See ATCA Starts to Rumble and ATCA Finds Its Way.)
Panelists also discussed some trends in broadband access, confirming a trend towards packet-based DSLAMs as opposed to the traditional ATM variety. That's being driven by the carriers' deployment of packet-based networks, but it's also come about because equipment providers "realized they can offer much cheaper DSLAMs" that way, says Imran Hajimusa, a director of marketing for Infineon Technologies AG (NYSE/Frankfurt: IFX). (See Chip Firms Target IP DSLAMs.)
Here, too, systems designs are becoming more dependent on merchant chips. The increased importance of video and other real-time traffic means traffic management could become a bigger part of the access picture. Equipment vendors face a tradeoff between density and increased processing, "sometimes scratching their heads over what a good balance is," Jankauskas says.
— Craig Matsumoto, Senior Editor, Light Reading