Celox Networks announced its first product today: the Celox SCx 192, a switch designed to manage IP subscriber services. It also announced its first trial customer: AT&T Corp. (NYSE: T).

Until now, Celox has kept its product plans quiet. In an industry where some companies make an announcement every time their CEO burps, it’s amazing that Celox has managed to get its IP services platform into customer trials and keep it quiet. At the same time, the company has managed to raise $68 million in a second round of funding while boosting its staff to 282, with facilities in St. Louis, Hudson, Mass., and India (see Celox Networks Nets $68 Million).

According to Hugh Kelly, VP of marketing, the SCx 192 supports up to six million simultaneous users with interfaces ranging from DS3 (45 Mbit/s) to OC 192 (10 Gbit/s) with an 80-Gbit/s switch fabric. This is in stark contrast to other companies marketing platforms for IP services, which include CoSine Communications Inc., Lucent Technologies Inc. (NYSE: LU), Nortel Networks Corp. (NYSE/Toronto: NT), and Redback Networks Inc. (Nasdaq: RBAK). For example, Cosine supports roughly 100,000 simultaneous connections and interfaces up to OC3 (155 Mbit/s) with a 24-Gbit/s fabric.

"We weren’t the first to market with a product," says Kelly. "So we wanted to make sure we had a customer and our core technology finished before we announced anything."

But some of Celox’s competitors are skeptical. "I don’t see any practical way that they could have built something 60 times more scaleable than our product," says David Messina, director of product marketing for Cosine. "Sometimes competitors come up with unrealistic claims, because the bar has already been set and their only opportunity is to up their marketing story.”

But certainly a deal with AT&T, one of the largest interexchange carriers around, is a sign that the Celox is for real. Trials are expected to begin within the next few weeks. And even though acceptance and full deployment could still be a long way off, Celox says that it has other customers lined up for trials and expects to be shipping products commercially beginning this summer.

“The AT&T commitment is huge for a startup and, to me, validates what they're doing,” says Scott Clavenna, president of PointEast Research LLC and head of Light Reading research. “Impressive stuff.”

Just getting into AT&T’s test bed is a huge achievement, but it’s not completely surprising. Before joining the company last January, Celox’s president and CEO Kent Mathy worked at AT&T for over 18 years. As Vice President of AT&T Global Services, Mathy developed and directed sales strategy to support a $2 billion market segment.

“If they were to get into AT&T’s network I’m sure it would be because of the obvious ties they have with the organization,” says Cosine's Messina.

But both Celox and AT&T deny any sort of favoritism. “Sure it may have helped us get in the door,” says Kelly. “But it’s just like being the movie director's kid. You may get the audition, but you wont get the part if you can’t act.”

What’s more, neither AT&T nor any of its employees has taken or bought any pre-IPO shares in the company, according to Kelly. “We don’t give out any friends-and-family shares,” he says. “We saw the questions that it raised when Cosine filed for its IPO. It just makes things too messy, and we didn’t see the advantage for us.”

How is Celox able to achieve the scale and performance that it claims? According to Kelly, the secret lies in the architecture and the special BitRipper network processor, a group of six custom ASICs (application-specific integrated circuits) that perform packet inspection, reassembly, reformatting, and scheduling of user flows at wire speeds up to OC192.

This architecture allows service providers to add processing capacity independently of interface capacity. This is important for service providers that already have sufficient network connections to their customers but would like to offer additional IP services over those connections, which requires more processing power. As more services are sold to subscribers, the provider can add processing engines to increase the capacity, says Kelly.

“Carriers don’t know what their networks are going to look like in the next five years,” he says. “And we wanted to give them something that would allow them to be as flexible as they can be.”

-- Marguerite Reardon, senior editor, Light Reading, http://www.lightreading.com