Ciena 1, Cisco 0
Instead, optical competitor Ciena developed an O-E-O version of “optical” switching that relied on electrical processing. Their Lightera-acquired product, rebranded as CoreDirector, was actually a bandwidth manager, which gave traditional cross-connect vendors such as Tellabs and Alcatel fits too. Such O-E-O-based bandwidth management was viewed as next generation cross-connect technology.
Cisco’s optical team, however, believed Ciena could be challenged with the Internet giant’s planned O-E-O entry, the ONS 15600, an advanced O-E-O product that appeared, on paper, to be a CoreDirector on steroids. Unfortunately, Cisco’s new optical leader who replaced Carl Russo, was lukewarm to the idea, and the ‘454’-inspired product never really took off. This failure marked Cisco’s second loss to Ciena. Attribute this market misstep to Jayshree Ullal, who erroneously believed that SONET technologies, hence O-E-O-based switching, would be obsolete by 2004.
Ciena 2, Cisco 0
What Happened and Why Was Cisco’s Leadership Wrong About Optical Trends?
A Fiber Optics News center page story, Beleaguered Optical Switching Industry Changes on the Fly, published on March 4, 2002, confirmed the wisdom of startup Cerent’s original 1997 choice of using conventional electronic switching chips versus trying to incorporate optical switching technologies in its Cerent 454 and subsequent product offerings.
Monterey Networks, acquired at same time as Cerent, for its MEMS-based O-O-O technology, proved more challenging. MEMS-based optical switching was a premature technology in 2000, the associated product development was bogged down within the Monterey team, and Cisco missed AT&T’s deadlines for that company to consider evaluating it.
Greg Nehib, a former Monterey employee, who remains at Cisco to this day, echoes those concerns, “The 15900 [Wavelength Router or optical switching product] was cancelled after Cisco missed AT&T's lab entry date. AT&T ultimately told the 15900 team that they lost their spot in the lab.”
As Greg and I discussed in 2015 about AT&T wanting such advanced technology for their long distance network, Greg noted, “I know the AT&T guys expected Cisco to ‘double down’ and beg forgiveness, but like you stated, it was a tough market and an easy project to bail on under the circumstances.”
Most of the 15900 development team joined the Cerent-inspired 15600 project team immediately after the 15900 was canned. If you can’t beat them with an O-O-O solution, join the O-E-O fray. The expanded “Cerent” optical team, infused with Monterey engineers, accelerated their development of the ONS 15600 in order to take on Ciena’s popular CoreDirector.
An O-E-O battle would soon heat up.
AT&T had Pushed for Optical Switching
AT&T’s technology gurus may have been the exception when they latched onto Monterey’s O-O-O offering, but the majority of service providers were correct: This form of optical switching was too early.
“The market has really been slow to take off,” said Carl Russo, group vice president of Cisco's Optical Networking unit, in April 2001, when Cisco announced the End-of-Life status for its 15900. “Right now, [service providers'] biggest concern is not adding capacity or functionality to the core networks as much as it is getting customers onto the existing network. That has a downward effect on the space the 15900 plays in, while at the same time having an upward effect on the metropolitan space.” [1]
Gary Smith, Ciena’s CEO, in October 2001, said of Cisco, “I think they’ve been successful with Cerent and I think they will continue to be successful if they stay at the edge of the network, that seems to be their forte.” [2]
By the spring of 2002, components and optical subsystems suppliers – Corning, OMM, Onix Microsystems, C Speed – which expected big returns from the O-O-O optical switching market, were earnestly looking for a new source of business. The more attractive optical cross-connect (OXC) and reconfigurable optical add-drop multiplexer (ROADM) markets had not yet gained traction across the optical market and next-generation SONET/SDH, whereas multi-service provisioning platforms (MSPPs) continued to be the optical platforms of choice.
And there’s the rub for optical switching products. “So it turns out that the all-optical cross-connect is a real product and is likely to grow into a significant market five or six years down the road,” Gasman says. “But that’s not quick enough if you’re a small startup making optical switching subsystems.” Nor was it quick enough for a large company like Cisco who killed their MEMS-based optical switching system in April 2001 after acquiring it from Monterey in August 1999.
Ultimately, Cisco would position its Cerent 454 to address two adjacent markets by splitting it into two products line. This divergence allowed the ‘454’ brand to be presented as the solution for both (what became) traditional MSPP support and the eventual demand for ROADMs (called MSTPs, or multi-service transport platforms, by Cisco). This strategy allowed Cisco to evolve an all-optical platform and leverage its optical experts based in Europe (the acquired Pirelli team located in Monza, Italy).
In the switching arena, however, what took off, instead of the all-optical switch, was the O-E-O switch. Service providers liked the grooming capability of the O-E-O switches, which is precisely the thing an all-optical switch cannot do. So Lucent and Corvis were the only equipment vendors of any size that sold the O-O-O or all-optical switches. Consequently, they were excluded form the larger O-E-O switching market. Ciena dominated this segment at the time and CIR expected the company to be increasingly challenged by competing vendors, such as Tellium and Altamar, who offered newer O-E-O switches. But this did not happen either.
Cisco’s entry, its ONS 15600, a high-end ‘454’ with expanded O-E-O switching capability, tried to challenge Ciena, but failed to gain significant traction in the O-E-O space. Ciena’s CoreDirector product, acquired from a Vinod Khosla-funded startup, crushed Cisco’s offering, among others.
Even so, Cisco’s optical team nipped at Ciena’s heels bringing in, on average, less than $10 million annually, primarily in support of ‘454’ sales or the chance to grow the metro optical transport business. Greg Nehib recalls, “Even though we were able to gain some high profile customers like Sprint, DISA, and LGN the product never achieved even modest feature parity with Lucent or Ciena.”
Why was that?
Greg notes, “The original development plan that was approved and funded by Terry Brown was not followed through with in the Jayshree era since she thought TDM would die a quick death in 2004. It's hard to say if that was a good choice as Ciena’s CoreDirector became a billion dollar product in the years that followed, with no competition. The 15600, by comparison, made [only] $100M in its lifetime before it was manufacture-discontinued in 2012.”
On the strength of Ciena’s CoreDirector offering, the O-E-O market thrived during the mid-2000s, as did Cerent’s ONS 15454 and its product extensions that remained true to the optical transport market instead of the optical switching market.
But not all of Cisco’s development efforts were in vain. Maybe patience is indeed a virtue, as Cisco, according to Greg, found success with the progeny of the ONS-15454 inspired ONS15600, “The 15600 chassis and system architecture was considered for re-use in the platform that is now the ASR9000 and the newer NCS4000, due to its split backplane and the ability to route/switch multiple traffic types.”
That’s an example of great engineering and persistence: the Cerent 454 architecture has always been viewed as a star of the telecom firmament!
[1] As reported by Jim Duffy in IT World, Cisco kills the Wavelength Router, April 6, 2001, online.
[2] Despite Stock Slide, Ciena Charges Onward, Fiber Optics News, October 1, 2001, p.8
[3] For more on AT&T and Doug Juers’ stewardship of that service provider for Cisco, please see my May 6, 2016 BLOG post.