Chip Roberson became Cerent’s third employee after the Fiberlane splinter in early 1998. He is proud of his role as the initial software coder at Cerent, so proud that he subsequently set up a Wikipedia article for Cerent Corporation on the web. Most Cerent alumni are proud of their contributions as well; gratified over the (R)evolution of the optical transport market they brought to the telecommunications sector during the late 1990s and early 2000s.
Serendipity Plays a Role in One’s Career
Chip’s career began in telecom with Alcatel in Raleigh, North Carolina, in 1990. He worked there for five years but became restless. By 1995, he arrived at the conclusion, “I can go anywhere.” The environment at Alcatel had became toxic, “It was so political . . . I felt underutilized.”
Serendipity Plays a Role in One’s Career
Chip’s career began in telecom with Alcatel in Raleigh, North Carolina, in 1990. He worked there for five years but became restless. By 1995, he arrived at the conclusion, “I can go anywhere.” The environment at Alcatel had became toxic, “It was so political . . . I felt underutilized.”
Chip soon entertained two job offers brought by headhunters: one from DSC, which meant employment with the Optilink team in Petaluma at the time, and the second one from Fujitsu at their Plano, Texas facility. Chip ultimately chose Optilink-DSC.
At Alcatel, Chip had worked on a team developing the company’s 150 Mbps optical transport product, a standards-based product that ultimately played a leading role in demonstrating SONET interoperability years later.
By the time Alcatel and its 1603 SM SONET multiplexer [1] participated in that ten-vendor interoperability demonstration at Supercomm ‘97, Chip had already departed Alcatel, spent two years at Digital Switch Corporation (DSC), and had just joined Fiberlane in February of that year.
Meanwhile, the much-publicized May 1997 interoperability event was coordinated by the SONET Interoperability Forum (SIF) on the Supercomm tradeshow floor. The Alcatel 1603 SM SONET multiplexer performed the role as “gateway,” responsible for the overall network management of the connected nine other OC-3 vendors. This key technical support included providing overhead communication (DCC) interfaces among all the demonstrating vendors for the OC-3 path-protected (UPSR) configuration. [2]
It was during his stint at DSC that Chip’s path crossed those of Ajaib Bhadare and Paul Elliott. Both engineers left an impression on Chip, and when Ajaib and Paul left DSC to pursue other interests, Chip was devastated. However, he soon saw an advertisement for software engineers for a new company down the street, Fiberlane.
At Alcatel, Chip had worked on a team developing the company’s 150 Mbps optical transport product, a standards-based product that ultimately played a leading role in demonstrating SONET interoperability years later.
By the time Alcatel and its 1603 SM SONET multiplexer [1] participated in that ten-vendor interoperability demonstration at Supercomm ‘97, Chip had already departed Alcatel, spent two years at Digital Switch Corporation (DSC), and had just joined Fiberlane in February of that year.
Meanwhile, the much-publicized May 1997 interoperability event was coordinated by the SONET Interoperability Forum (SIF) on the Supercomm tradeshow floor. The Alcatel 1603 SM SONET multiplexer performed the role as “gateway,” responsible for the overall network management of the connected nine other OC-3 vendors. This key technical support included providing overhead communication (DCC) interfaces among all the demonstrating vendors for the OC-3 path-protected (UPSR) configuration. [2]
It was during his stint at DSC that Chip’s path crossed those of Ajaib Bhadare and Paul Elliott. Both engineers left an impression on Chip, and when Ajaib and Paul left DSC to pursue other interests, Chip was devastated. However, he soon saw an advertisement for software engineers for a new company down the street, Fiberlane.
Chip responded to Ajaib’s request for engineering talent almost immediately. He interviewed and joined Ajaib and Paul (or Mr. Patent, as he was affectionately known), on Valentine’s Day, 1997. It was a technology-based love affair.
Chip said he clearly recalls his one-hour interview at 1318 Redwood Way in Petaluma with Ajaib. Chip snuck in Fiberlane’s back door, since everybody in telecom new everybody else in telecom back then, especially among the close-knit telecom community in the North Bay area.
The interview began with a skeptical software engineer and ended with Chip excited over the opportunity to work for this latest startup in Telecom Valley. Chip was convinced that available new technology [3] would enable something really special to be done in the optical transport space. In fact, an all-optical product could be realized, something not even being contemplated by the large established players in the optical transport space. Chip saw things come full circle: one day IP would soon arrive in the home. And it has.
Chip said he clearly recalls his one-hour interview at 1318 Redwood Way in Petaluma with Ajaib. Chip snuck in Fiberlane’s back door, since everybody in telecom new everybody else in telecom back then, especially among the close-knit telecom community in the North Bay area.
The interview began with a skeptical software engineer and ended with Chip excited over the opportunity to work for this latest startup in Telecom Valley. Chip was convinced that available new technology [3] would enable something really special to be done in the optical transport space. In fact, an all-optical product could be realized, something not even being contemplated by the large established players in the optical transport space. Chip saw things come full circle: one day IP would soon arrive in the home. And it has.
Fiberlane to Cerent
After a short time at Fiberlane, Chip felt that his newly adopted company was actually comprised of three separate teams holding three worldviews and working towards three different products: The SONET multiplexer engineers in Petaluma, the IP-focused data team toiling in Mountain View, and a network management appendage in Canada’s Vancouver area.
“In the beginning days, the guys in the South Bay really wanted to build a Cisco-killer, the next big router, big iron.” Chip believed his Silicon Valley colleagues were impatient to get the SONET multiplexer portion of the product completed. They said, “Let’s get past this telecom box as soon as we can.”
Looking northward, his Canadian colleagues were “network management guys.” Chip recalls, “They said your platform is a really good one for showing what our management system can do.”
After a short time at Fiberlane, Chip felt that his newly adopted company was actually comprised of three separate teams holding three worldviews and working towards three different products: The SONET multiplexer engineers in Petaluma, the IP-focused data team toiling in Mountain View, and a network management appendage in Canada’s Vancouver area.
“In the beginning days, the guys in the South Bay really wanted to build a Cisco-killer, the next big router, big iron.” Chip believed his Silicon Valley colleagues were impatient to get the SONET multiplexer portion of the product completed. They said, “Let’s get past this telecom box as soon as we can.”
Looking northward, his Canadian colleagues were “network management guys.” Chip recalls, “They said your platform is a really good one for showing what our management system can do.”
Chip summarizes the Fiberlane situation, “We basically had three different teams developing three different products. We weren’t coherent. Not everyone was trying to do the same thing. [In Petaluma,] we were trying to build the Cerent box.”
After the eventual split with Mountain View and Vancouver, it felt like our Board [4] “had cut the tethers.” Chip adds, “We just took off in Petaluma. I felt like I was a runner and someone had strings tied to me and I just couldn’t move forward. After the strings were cut, we launched forward.”
Pixar’s movie, UP, provides an uplifting visual analogy. Once the ‘separation from Earth’ was made, the ‘balloons of creativity’ lifted the Petaluma engineers of Fiberlane above it all and they entered the stratosphere. Here they could focus on the idea and vision of one product and a shared worldview, a product that would become known as the Cerent 454.
The Petaluma team didn’t understand why the Mountain View team had wanted to build the so-called Cisco-killer. To them, it seemed that Juniper had a better chance at beating Cisco than Fiberlane ever would have. On this score, Chip notes, “While I was at Fiberlane it felt like I was pushing on a noodle to make any headway. Our interests just weren’t aligned.”
Chip took comfort in Vinod Khosla’s overarching vision, which Vinod reportedly said when he decided to fund Fiberlane in the first place, “We had @Home, and we had Juniper, and we had nothing in the middle of our portfolio. We saw there was a gap. We knew we needed that next generation network. We needed that piece.”
Indeed. An innovative optical transport product was needed to deal with the tsunami of bandwidth coming at the network. Chip notes, “That was part of the early division, when Jay [Sethuram] and Raj [Singh] left. We’re not doing Jay’s version of the product. We’re doing Ajaib and Paul’s version of the product.”
After the eventual split with Mountain View and Vancouver, it felt like our Board [4] “had cut the tethers.” Chip adds, “We just took off in Petaluma. I felt like I was a runner and someone had strings tied to me and I just couldn’t move forward. After the strings were cut, we launched forward.”
Pixar’s movie, UP, provides an uplifting visual analogy. Once the ‘separation from Earth’ was made, the ‘balloons of creativity’ lifted the Petaluma engineers of Fiberlane above it all and they entered the stratosphere. Here they could focus on the idea and vision of one product and a shared worldview, a product that would become known as the Cerent 454.
The Petaluma team didn’t understand why the Mountain View team had wanted to build the so-called Cisco-killer. To them, it seemed that Juniper had a better chance at beating Cisco than Fiberlane ever would have. On this score, Chip notes, “While I was at Fiberlane it felt like I was pushing on a noodle to make any headway. Our interests just weren’t aligned.”
Chip took comfort in Vinod Khosla’s overarching vision, which Vinod reportedly said when he decided to fund Fiberlane in the first place, “We had @Home, and we had Juniper, and we had nothing in the middle of our portfolio. We saw there was a gap. We knew we needed that next generation network. We needed that piece.”
Indeed. An innovative optical transport product was needed to deal with the tsunami of bandwidth coming at the network. Chip notes, “That was part of the early division, when Jay [Sethuram] and Raj [Singh] left. We’re not doing Jay’s version of the product. We’re doing Ajaib and Paul’s version of the product.”
Chip felt that Jay was talking about the Enterprise play versus the Service Provider play in the realization of the product. The Mountain View team was focused on moving to the Enterprise space to take on Cisco. The Petaluma team, on the other hand, wanted to move to the Service Provider space and go after, initially, the smaller telco customers to build the revenue ramp for their vision of the product. Vinod ultimately became sympathetic to this case.
“Two different paths, two different perspectives, two different markets. I think Vinod saw that,” said Chip. “Service Providers needed to build out the heart of the network, [the transport space] . . . You can see how many different visions were in this company from the very beginning. One by one they had to be lopped off or modified.”
The notion of distilling the vision is apt. Chip reflected on one of his advisors, Jim Levy, Activision’s founder: “The first rule of a start-up is to survive long enough to figure out how to be successful. And in some regards, Cerent had to do that. We started out with multiple visions, and that distilling down allows you to figure out what it is.”
“Terry [Brown] and Tom [Corker] brought customers to the table very quickly,” Chip recalls. “We could have built an amazing product [as Fiberlane], but it may have not have been what customers wanted.”
That is one secret of success for any upstart startup: Build a product that customers have input to, that operators want. And that ended up being the Cerent 454 that emerged out of the engineering labs of Cerent in Petaluma.
I learned a lot about following Ajaib and then working for Gary Baldwin, our software director that started for Ajaib in the fall of 1998, “Gary brought the one thing that was most valuable, and that was discipline, that took a start-up team and made it into an organization that was capable of producing a real product. A ‘productization’ that has to happen, that has to be reliable, requires great discipline.”
“Two different paths, two different perspectives, two different markets. I think Vinod saw that,” said Chip. “Service Providers needed to build out the heart of the network, [the transport space] . . . You can see how many different visions were in this company from the very beginning. One by one they had to be lopped off or modified.”
The notion of distilling the vision is apt. Chip reflected on one of his advisors, Jim Levy, Activision’s founder: “The first rule of a start-up is to survive long enough to figure out how to be successful. And in some regards, Cerent had to do that. We started out with multiple visions, and that distilling down allows you to figure out what it is.”
“Terry [Brown] and Tom [Corker] brought customers to the table very quickly,” Chip recalls. “We could have built an amazing product [as Fiberlane], but it may have not have been what customers wanted.”
That is one secret of success for any upstart startup: Build a product that customers have input to, that operators want. And that ended up being the Cerent 454 that emerged out of the engineering labs of Cerent in Petaluma.
I learned a lot about following Ajaib and then working for Gary Baldwin, our software director that started for Ajaib in the fall of 1998, “Gary brought the one thing that was most valuable, and that was discipline, that took a start-up team and made it into an organization that was capable of producing a real product. A ‘productization’ that has to happen, that has to be reliable, requires great discipline.”
Notes:
[1] Alcatel’s 1603 SM (SONET Multiplexer) was according to the company, a “third-generation SONET system” that could be configured to support a wide range of applications including OC-3, OC-12 or OC-48 capacities; linear or ring transport; optical hubbing; or a combination of traffic payloads. Subsequently, the 1603 SMX (SONET Multi-Application Network Element) was a “multiservice SONET network element with assorted interfaces for deployment agility, full SONET standards compliance, and up to 16 OC-3 single-shelf drop capability.” Network efficiency increased in this version of the 1603 with VT1.5 or STS-1 level grooming and STS-level bandwidth management across the entire OC-n payload. In addition, an M13 transmultiplexer provided DS3 to VT payload mapping, a testament to the popularity of Cerent’s “Transmux” card option for its own ‘454.’
[1] Alcatel’s 1603 SM (SONET Multiplexer) was according to the company, a “third-generation SONET system” that could be configured to support a wide range of applications including OC-3, OC-12 or OC-48 capacities; linear or ring transport; optical hubbing; or a combination of traffic payloads. Subsequently, the 1603 SMX (SONET Multi-Application Network Element) was a “multiservice SONET network element with assorted interfaces for deployment agility, full SONET standards compliance, and up to 16 OC-3 single-shelf drop capability.” Network efficiency increased in this version of the 1603 with VT1.5 or STS-1 level grooming and STS-level bandwidth management across the entire OC-n payload. In addition, an M13 transmultiplexer provided DS3 to VT payload mapping, a testament to the popularity of Cerent’s “Transmux” card option for its own ‘454.’
[2] DCC stands for Data Communications Channel while UPSR is an acronym for Unidirectional Path Switched Ring. According to the SONET standard, the DCC is found in the section layer, comprised of three bytes (D1, D2, D3) allocated in STS-1 of an STS-N signal. The DCC provides support for section layer data communications. These three allocated bytes are treated as a single 192 kbps data channel for the transmission of alarms, maintenance details, control information, and administration details as well as other network element communication needs. In the SONET line layer, nine bytes (D4 through D12) provide three more 192 kbps data channels, for an aggregate 576 kbps, for similar purposes. Cerent’s software innovators enabled the ‘454’ to balance data flow over these DCC channels using a load balancing algorithm. This algorithm chose a DCC for packet transport by considering packet size and DCC utilization.
In terms of UPSR, or path protection, each ‘454’ node bridges its transmit information across both the working and protect lines. When traffic is switched from a bad line, only the receiving node performs a switch. The APS channel (which is carried in the K1 and K2 bytes of the signal on the protection line) is used to indicate the local switch action and the mode of operation. Path protection is the default mode for 1+1 protection groups in the ‘454.’ For more, please visit Cisco’s ONS 15454 webpage.
In terms of UPSR, or path protection, each ‘454’ node bridges its transmit information across both the working and protect lines. When traffic is switched from a bad line, only the receiving node performs a switch. The APS channel (which is carried in the K1 and K2 bytes of the signal on the protection line) is used to indicate the local switch action and the mode of operation. Path protection is the default mode for 1+1 protection groups in the ‘454.’ For more, please visit Cisco’s ONS 15454 webpage.
[3] In The Upstart Startup, we review a number of key contributions Chip made to the ‘454’ product, including, development of the foundational software, which led to the first software release of the ‘454;’ adoption of Wind River’s real-time operating system, VxWorks, that came with a built-in TCP/IP stack; leveraging of the linear addressing capability of the Motorola chip set, as opposed to Intel’s more difficult offering; and, sorting out custom-built ASIC interactions during 1999.
[4] Two of the Cerent VCs interviewed in 2013, Promod Haque and Don Green, agreed that such a separation of the Cerent team (Fiberlane-Petaluma) from the Siara Systems team (Fiberlane-Mountain View) was key to Cerent’s eventual success and astronomical valuation in 1999 as the upstart startup was targeted by Cisco Systems.
[4] Two of the Cerent VCs interviewed in 2013, Promod Haque and Don Green, agreed that such a separation of the Cerent team (Fiberlane-Petaluma) from the Siara Systems team (Fiberlane-Mountain View) was key to Cerent’s eventual success and astronomical valuation in 1999 as the upstart startup was targeted by Cisco Systems.