Many industry analysts and optical networking observers during the 1990s pronounced that deployment of traditional SONET equipment was no way to build an optical transport infrastructure.
One industry pundit, George Gilder, wrote, “To light up a . . . thirteen thousand mile network with a full complement of WDM wavelengths would cost $250 million for the first SONET lambda of 10 gigabits per second and perhaps $80 million each for the next seven. Lambda number nine would cost another $250 million. To light just one fiber thread with the 160 lambdas that Nortel offered at the end of 1999 would cost some $16 billion.”
Here, Gilder calculated the cost of building nationwide networks with traditional SONET, OC-192 equipment from Nortel Networks. QWEST and Williams were two such companies using OC-192 at the time.
He added, “And that would not even include the costs of real estate, cross-connects, add-drop muxes, and the other gear needed actually to deliver services to customers along the network. WDM promised the transformations of the flood, but [traditional] SONET would strangle it at the source.”
Indeed, integrated DWDM would help with the economics, but Gilder did not look at the emerging Multi-service Provisioning Platform (MSPP) offerings pioneered by Cerent. This could be because his book was written in 1999 (published in 2000) and he was unaware of the breakthrough Cerent introduced that year, the year Williams began using the Cerent 454 as a bandwidth manager and QWEST began evaluating it for terminating Nortel’s high-priced OC-192 links.
The use of 4F- and 2F-rings ensured reliability and the new mesh architecture that Cerent offered supported IP traffic even better than the ring architectures for many reasons, including that capability to deploy capacity only where needed and switch traffic locally within the MSPP nodes.
Nortel (and Lucent too) had to eventually scrap their previously successful legacy SONET platforms after Cerent introduced its much more cost-effective MSPP that allowed two transport layers to exist simultaneously. Cerent was able to integrate a viable fiber optic circuit connection, using the SONET standard, but it also embraced the more cost-effective transport of Internet data using layer 2 switching and the carriage of Ethernet/IP. Not only that, but Cerent also helped service providers, both new CLECs and eventually established telephone companies, move Ethernet service offerings from an idea to a tariffed service. Cerent was actively promoting this as a transport service for the high-margin business customers courted by the local telecom providers.
What the Cerent 454 product did so well was to address the feature functionality that telephone companies were asking for. However, in all of Nortel’s arrogance about knowing better what to do than any customer, the optical transport giant was building their optical transport boxes according to bit rate—OC-3, OC-12, OC-48, and OC-192—all in different physical sizes and disparate mechanical packages. It was a “design by bit rate” mentality and an “organization by divisions,” with armies of engineers designing bit rate-optimized boxes. Rapid bandwidth growth during 1997 and 1998 meant that an OC-3 deployment was quickly obsolete and then an OC-12 replacement was needed and/or an OC-48 upgrade. Customers were tired of forklift upgrades to grow their optical transport networks. Replacing one box with another box was getting stale.
One industry pundit, George Gilder, wrote, “To light up a . . . thirteen thousand mile network with a full complement of WDM wavelengths would cost $250 million for the first SONET lambda of 10 gigabits per second and perhaps $80 million each for the next seven. Lambda number nine would cost another $250 million. To light just one fiber thread with the 160 lambdas that Nortel offered at the end of 1999 would cost some $16 billion.”
Here, Gilder calculated the cost of building nationwide networks with traditional SONET, OC-192 equipment from Nortel Networks. QWEST and Williams were two such companies using OC-192 at the time.
He added, “And that would not even include the costs of real estate, cross-connects, add-drop muxes, and the other gear needed actually to deliver services to customers along the network. WDM promised the transformations of the flood, but [traditional] SONET would strangle it at the source.”
Indeed, integrated DWDM would help with the economics, but Gilder did not look at the emerging Multi-service Provisioning Platform (MSPP) offerings pioneered by Cerent. This could be because his book was written in 1999 (published in 2000) and he was unaware of the breakthrough Cerent introduced that year, the year Williams began using the Cerent 454 as a bandwidth manager and QWEST began evaluating it for terminating Nortel’s high-priced OC-192 links.
The use of 4F- and 2F-rings ensured reliability and the new mesh architecture that Cerent offered supported IP traffic even better than the ring architectures for many reasons, including that capability to deploy capacity only where needed and switch traffic locally within the MSPP nodes.
Nortel (and Lucent too) had to eventually scrap their previously successful legacy SONET platforms after Cerent introduced its much more cost-effective MSPP that allowed two transport layers to exist simultaneously. Cerent was able to integrate a viable fiber optic circuit connection, using the SONET standard, but it also embraced the more cost-effective transport of Internet data using layer 2 switching and the carriage of Ethernet/IP. Not only that, but Cerent also helped service providers, both new CLECs and eventually established telephone companies, move Ethernet service offerings from an idea to a tariffed service. Cerent was actively promoting this as a transport service for the high-margin business customers courted by the local telecom providers.
What the Cerent 454 product did so well was to address the feature functionality that telephone companies were asking for. However, in all of Nortel’s arrogance about knowing better what to do than any customer, the optical transport giant was building their optical transport boxes according to bit rate—OC-3, OC-12, OC-48, and OC-192—all in different physical sizes and disparate mechanical packages. It was a “design by bit rate” mentality and an “organization by divisions,” with armies of engineers designing bit rate-optimized boxes. Rapid bandwidth growth during 1997 and 1998 meant that an OC-3 deployment was quickly obsolete and then an OC-12 replacement was needed and/or an OC-48 upgrade. Customers were tired of forklift upgrades to grow their optical transport networks. Replacing one box with another box was getting stale.
Furthermore, each SONET box supported one or two service interfaces and a limited amount of topological configurations. For example, Nortel’s industry-leading S/DMS TransportNode OC-48 supported DS3s and STS-1s in linear and two-fiber BLSR configurations, the very configuration that U S West deployed since 1995 in its key metropolitan areas.
But that was it. Nortel’s OC-48 box could not support other bit rates such as OC-12 on the line side or other service interfaces such as DS1s, at least not without a different box altogether or adjunct boxes, making the solution more costly, power hungry, and space inefficient. Add to the fact that business customers wanted to connect their campuses with secure transport using Ethernet or ATM, this legacy SONET gear from Nortel was fast becoming obsolete; it couldn’t support the burgeoning demand of data traffic since it had been designed at the outset for the optimal transport of voice traffic.
By 1998, Nortel was living off its past successes and not innovating in the metropolitan optical transport space. The company had moved its investment dollars to the long-haul applications with its industry-leading 10 Gb/s OC-192 system. It had all but abandoned the optical space in the metro, at least from a technological perspective, and had allowed Fujitsu to regain dominance in U S West, for example.
But this was Cerent’s time. This upstart startup would ultimately knock the Fujitsu monopoly out of the metropolitan space within U S West, as Fujitsu was caught off guard by Cerent’s technological innovation that took hold in many of this Baby Bell’s offices in 2000.
This carrier class product was presented to an expanding market by a world-class sales organization. As one former Cisco insider, Jayshree Ullal, commented about the Cerent 454 and its band of intrepid entrepreneurs that took on the Big Iron suppliers, “I’ve had so many people tell me I have a ‘454’ or I use your MSPP platform, so the branding and rapid adoption of that platform, really rendered the legacy SONET market obsolete and those legacy vendors flat-footed. That competitive advantage gave Cerent the ability to have one of the fastest meteoric rises in telecom history. It was zero to a billion run rate in three years. It definitely goes into the history books.”
[1] Slides included in this post used in R.K. Koslowsky's SSU Presentation on the Upstart Startup.
By 1998, Nortel was living off its past successes and not innovating in the metropolitan optical transport space. The company had moved its investment dollars to the long-haul applications with its industry-leading 10 Gb/s OC-192 system. It had all but abandoned the optical space in the metro, at least from a technological perspective, and had allowed Fujitsu to regain dominance in U S West, for example.
But this was Cerent’s time. This upstart startup would ultimately knock the Fujitsu monopoly out of the metropolitan space within U S West, as Fujitsu was caught off guard by Cerent’s technological innovation that took hold in many of this Baby Bell’s offices in 2000.
This carrier class product was presented to an expanding market by a world-class sales organization. As one former Cisco insider, Jayshree Ullal, commented about the Cerent 454 and its band of intrepid entrepreneurs that took on the Big Iron suppliers, “I’ve had so many people tell me I have a ‘454’ or I use your MSPP platform, so the branding and rapid adoption of that platform, really rendered the legacy SONET market obsolete and those legacy vendors flat-footed. That competitive advantage gave Cerent the ability to have one of the fastest meteoric rises in telecom history. It was zero to a billion run rate in three years. It definitely goes into the history books.”
[1] Slides included in this post used in R.K. Koslowsky's SSU Presentation on the Upstart Startup.