Plenty of startups are trying to build a terabit router, one that will outgun Cisco Systems, Inc. www.cisco.com and Juniper Networks, Inc. www.juniper.net, solve the capacity crunch in Internet backbones and make them billions of bucks in the process (see Terabit Turmoil ). But one startup - Hyperchip, Inc. www.hyperchip.com - stands out from the rest, from two points of view.
First, it's aiming to create something much more than a bigger, faster, box. It's aiming to create the Internet equivalent of a Class 5 telephone switch, something that would sit at the edge of optical backbones and handle IP connections to tens of thousands of users. Hyperchip's developments would potentially replace entire ISP POPs (points of presence) and would have an aggregate capacity measured in - get this - petabits a second.
Second, Hyperchip is addressing this requirement in a totally different (some would say bizarre) way. It's devoted most of its efforts into adapting supercomputer hardware to deliver the scalability it requires. Software - considered the key to success by most terabit router vendors and users - seems to be of secondary importance to the Montreal based startup.
So, is Hyperchip onto something, or is it out to lunch?
On the hardware front, Hyperchip sounds credible, according to industry analysts and other router vendors. It says it can create a single logical router with thousands of ports by stringing together a series of crossbar switch fabrics. Unlike other vendors that have attempted to scale routers in this way, Hyperchip says it has embedded scheduling and control intelligence into each switch ASIC, which allows for a more scalable architecture.
"It sounds like they could be onto something that might be valuable," says Carl Blume, director of product marketing for Ironbridge Networks, Inc. www.ironbridgenetworks.com, a competing startup. "But in terms of building an overall product, software is the secret sauce."
That's where the potential problem lies. Hyperchip doesn't have much secret sauce. Instead of developing its own code from scratch, it's starting with free "Open IP Environment" software from Nortel Network www.nortel.com and adding its own extensions.
"A lot of vendors want to be macho," says Richard Norman, president and CTO of Hyperchip. "But why re-invent the wheel if you don't have to? We're using this opportunity as an advantage."
Opinions differ on whether Nortel's Open IP software is a suitable starting point. Critics point out that it's general purpose software, designed for a wide range of products including personal digital assistants. They also note that Nortel hasn't released its own core router - casting doubts on its software skills in this field. "If you get the software for free, you get what you pay for," says John Stewart, marketing engineer for Juniper Networks. "It just doesn't seem like a recipe for success to me."
This view is challenged by Raj Mehta, an analyst with Ryan Hankin Kent (RHK) www.rhk.com. "There are probably more high-end router vendors using this software than not," he says. "It's not so much that they're using open source code; it's really more about what they've added to enhance the software."
The question mark hanging over Hyperchip's software is likely to be the startup's biggest problem. Carriers will need convincing that its equipment can peer with existing Cisco and Juniper routers before they'll countenance trials, which are scheduled to start at the end of this year.
Still, maybe it won't come to that. If Hyperchip's hardware story stands up, it could get bought before it gets to market.
Light Reading is an online resource for optical networking information. Visit us at www.lightreading.com.
RELATED LINKS
