IP Fabrics Targets Intel Support
IP Fabrics Inc. was started in 2002 with the aim of creating a new programming language for network processors. The approach is intended to work with anybody's chips, but the company is starting out by targeting the Intel Corp. (Nasdaq: INTC) IXP2300 and IXP2800 high-end chips, says CEO and founder Glen Myers.
The 17-employee company is sailing into dangerously shallow waters, however, as the network processor industry hasn't quite taken off (see Network Processor Revival). This is probably why the company chose to target Intel first. Intel and Applied Micro Circuits Corp. (AMCC) (Nasdaq: AMCC) are the top sellers of network processors, and Intel in particular has organized an army of support staff around the products (see Intel's Net Processor Invasion).
Intel also helped fund the project. Intel Capital is among the investors that have ponied up $6 million to get IP Fabrics going. The other investors are all located in the Northwest: Frazier Technology Ventures, Ignition Partners, and Northwest Venture Associates.
IP Fabrics will have to compete against the software provided by the chip companies themselves, as well as against third-party programming from the likes of Future Software Ltd. (FutureSoft) -- acquired by Flextronics Corp. (Nasdaq: FLEX) -- , IP Infusion Inc., LVL7 Systems Inc., and Teja Technologies Inc. (also funded by Intel -- see Why Intel Loves Teja ).
But while most of those companies build programs for network processors, IP Fabrics has developed its own programming language, cleverly called the Packet Programming Language (PPL). The tactic defies the conventional belief that it's best to offer network processors that can be programmed in C, a language plenty of people already know.
There's also the fact that every network processor vendor, Intel included, already provides software. "I've yet to see a proprietary language from an [independent software vendor] have any success," says Bob Wheeler, an analyst with The Linley Group. "In my opinion, Intel's new C compiler will torpedo these guys."
Why take on this market, then? As network processors have spread into the market, it's become clear that programming the chips is a tough task -- so much so that Agere Systems Inc. (NYSE: AGR.A) is making a big deal out of a recent textbook extolling the company's network processor software (see Author Lauds Agere Net Processor).
Solving the software question involves more than making the chips speak C, IP Fabrics contends. "The world needs a completely different approach," Myers says.
IP Fabrics claims PPL is actually simpler than C and therefore presents a significant advantage over any Intel software. PPL is a "high-level" language, meaning it can describe tasks in much fewer lines of code than C requires. IP Fabrics has five beta-test customers -- "instantly recognizable names," Myers says -- who have adapted to PPL. "They can create an application in a day and get it running on the network processor. That used to be a two-year project," he says.
In part, that explains why Intel would invest in the company. Intel Capital seeks out startups that can help boost demand for Intel's chips, even in instances where the startup might compete with Intel. If PPL can make life easier for network processor users, it makes sense that Intel would be all for it. (Intel Capital representatives were attending the Consumer Electronics Show late last week and couldn't be reached for comment.) As the founder of software firm RadiSys Corp. (Nasdaq: RSYS), Myers got early exposure to the programming difficulties of network processors. After being fired as CEO in 2002, when RadiSys got pummeled by the telecom downturn, Myers set off with a handful of RadiSys employees to start IP Fabrics.
IP Fabrics's approach mimics Java, although Myers winces at PPL being called "Java for network processors." The company's software sits in a virtual machine atop the chip. The virtual machine acts as a translator, taking in PPL commands and executing them on the chip.
— Craig Matsumoto, Senior Editor, Light Reading
For further education, visit the archives of related Light Reading Webinars: