At the OFC 2015 show next month, the Rump Session is entitled, “Is it 'Game Over' for Hardware?” The discussion will be about “[s]ome industry trends like SDN appear[ing] to make optics hardware engineering and innovation irrelevant and shift innovation into software-based network control [while] [o]ther industry trends like [s]ilicon [p]hotonics appear[ing] to create great new opportunities for hardware innovation.” Of course, while around the world, all kinds of fantastic developments in fiber optic gear continually take place in laboratories, the really important focus should be on actual implementation of new solutions in the telecom space, which has historically tended to be driven by events or needs that were unpredictable in nature.
Just the specific network requirements of one large carrier can determine the success of a particular solution. In the same way, the decision of a single person on telecom policy, such as an FCC chairman, can make a big difference on a new type of device taking off.
While “[l]ow noise EDFAs enabled WDM,” it is uncertain there would have been much of a market today without Sprint’s unusually big necessity for ring capability on its 10G systems. Nortel initially lacked ring functionality on its OC-192 systems. Therefore, Ciena’s existence was enabled, and the rest is history.
The impact of aberrational changes on deployment of advanced technology has been seen in all areas of telecom. For the most part, Western Electric was the only developer and provider of network equipment within Ma Bell. In the 1970s, a Canadian supplier, Northern Telecom, revolutionized the central office space with digital solutions, and the Bell companies were adamant about getting them. Although AT&T, the corporation, was kicking and screaming (because after all, analog switches worked), it eventually relented and allowed the local carriers to install digital COs.
Obviously, in the 1980s, the big aberration was the breakup of Ma Bell, and new long-haul service providers had to build networks in a hurry – and in doing so, introduced single-mode fiber to the US. During this decade, AT&T itself ultimately moved in this new direction. Nevertheless, it is scarcely out of the question that in the '80s, massive deployment of other than multimode fiber would not have occurred in the US without divestiture by the main telco.
Up until the present, there have been several other unanticipated market and network conditions that have resulted in the migration to novel types of optical technology. We fully expect that history will continue to repeat itself in this way long into the future.
Getting back to the main premise of the OFC event, we have been very clear on our doubts about the extensive use of Si photonics for active gear. At least in the public market, SDN was overhyped right from the start. It may be enough to point out that in a little over three years, SDNCentral.com changed its name to SDxCentral.com.
We also take exception with the notion that the “[h]igh initial cost of 10G in the '90s was accepted by industry, while similar high initial cost of 100G has been widely criticized.” There was at least one major carrier in which Nortel provided a 10G frame, while only equipping with 2.5G capability because the customer did not believe at first it needed the higher capacity – let alone the additional cost. Also, while the Internet created legitimate demand for much greater amounts of bandwidth and the bubble resulted in irrationally large installations of OC-192, today there are hardly such catalysts helping to drive up volume and reduce cost.
Moreover, 100G is the first time in the optical space in which there was no VC money contributing to the development of these systems. This internal pressure on equipment vendors to get a return on investment in combination with margin pressures on the service providers has resulted in hypersensitivity over cost, which did not exist to such an extent in the ‘90s.
Another factor to consider is if the telecom industry adopted the correct mentality of moving in more of an all-optical approach, which would be the flattest and cheapest way to go forward, there would be plenty of innovation found in networks.
[written by Mark Lutkowitz]