In defiance of AT&T’s uniquely leviathan network involving layers of infrastructure built over many years, the service provider continues to promote further leading-edge concepts that would ostensibly change the fundamental nature of its cost structure as well as its spider web of operational entanglements. These major shifts have been projected to occur in an immense way a long way down the road, whether it was getting to all-IP by 2020 (somehow in 2012, all of the T1s would be expected to disappear and OTN would cease to becomes a reality), and in 2014, it projected that over 75% of its network would be software-centric by the end of the decade. AT&T’s latest scheme was introduced at OFC 2016, the "Open ROADM", and while also based on the notion of "software control," it brings in the idea of "open hardware" or Open Line Systems (OLS) in which Verizon, at the same conference, discredited the concept as a viable option for incumbents. In addition, the lion’s share of any company's well-established optical installed base would apparently not be affected by a move toward virtualization and white boxes promoted by what appears now to be a relatively small number of hardcore SDN advocates.
It is probably not an overstatement that Verizon leads the world in terms of technological advancements involving high-capacity, optical transmission in the public market. The carrier’s Director of Optical Transport Network Architecture, Design and Planning, Glenn Wellbrock, who is widely known as a straight-shooter, did not mince his words at either a Service Provider Summit panel or at the Rump Session at OFC on the matter of the pitfalls associated with this kind of disaggregation, especially in the metro, which AT&T is particularly targeting for "Open ROADM," a sector with far greater complexity in terms of engineering than in the long haul. He more than implied that OLS entailed the risk of blowing up the network based on the uncertainties associated with maintaining adequate long-term performance and reliability, even if an outside systems integrator took on the responsibility for putting the piece-parts together.
There is no question that it has been important for AT&T to push such narratives in order to persuade investors that its financial position will eventually improve with these actions. Yet, when pressed on the validity of these plans becoming a reality to the extent as promised, as a provider with the least amount of adaptability internationally, given its size and its age, it seems that it is forced to continually acknowledge that the focus is on greenfield applications, which by definition, precludes a massive amount of change.
There is not much evidence to suggest that AT&T has become in reality substantially less operationally driven based on the old Ma Bell model, which has always advocated keeping elements in the network indefinitely. In contrast, again, it is Verizon, despite its conservative stance on OLS, which has been quite a bit more innovative. For example, we are fairly certain that no engineer there would have been fired in hypothetically bending over backwards in the wireless backhaul, even a few years ago, by deploying a 100G system because the last thing that Verizon desires is to replicate the disaster that happened with AT&T in having an inadequate amount of capacity when it initially took on Apple’s iPhone.
Along the same way of thinking, AT&T still has a 10:1 split between 40G and 100G on its line-side installed base, given that in previous years the former was preferred because of its lower price and the large number of 40-gig interfaces on its routers and cross-connects. It would hardly be surprising if the operator puts off 400G in a similar way because of the standard operational considerations on the user end ultimately with 100-gig.
In returning to the topic of OLS, we expressed our reservations last October over such an approach, particularly involving the older users. We clearly did not count on the distressing willingness of several vendors to cooperate so fully in an arrangement (starting with the hyperscale data center firms) that could be so potentially damaging to their future businesses. While the old guard at "AT&T Network Systems," Alcatel, and "Northern Telecom," were not all necessarily made up of the sharpest knives in the drawer, they would have fought tooth and nail before engaging in such activities, regardless of marketplace conditions.
[written by Mark Lutkowitz]