At the OFC 2015 conference, one of the most highly respected network engineers in the business from Xtera Communications will be legitimately asking the question on a panel, “What on Earth is a ‘100G Metro System?’” Bill Szeto will be asserting that because 100G currently means a whole wavelength, there is nothing wrong serving the needs with just taking a 100-gig channel off of a ROADM (Reconfigurable Optical Add/Drop Multiplexer). In doing so, it would also take care of the potential problems with stranded capacity that would occur in mesh networks as we outlined in a previous blog article.
Bill will also argue that a so-called 100G metro system would provide no additional features/functionality. Also, there would be the avoidance of the issues today, such as with compatibility between different vendors' equipment at that data rate. Obviously, service providers would not want to get locked into a single vendor situation to hook up to the core.
Certainly, over the years, there have been problems with 10G systems from separate suppliers talking to one another – but not to the degree found with today’s more complex 100G gear including the use of proprietary FEC. Compatibility matters with 10G can be addressed by going into an OTN switch. In the past, at least theoretically, there were mid-span meets at that speed using SONET/SDH technology.
Given the underwhelming amount of expenditures on 100G by service providers for backhaul compared to the initial expectations, the ability of vendors to increasingly get away with hyping the notion of metro 100G is quite remarkable. The extremely premature nature of any type of discussion on this subject is bolstered by Bill’s remark to us that he is unaware of any contiguous 100-gig network that is available in the US -- much of which is because of the lack of equipment standardization.
There is universal agreement that the real requirement for 100G is for data center connections, which can be legitimately called metro applications, if they are within, say, 45 kilometers of one another. However, it is not a secret that at present, 100-gig ports do not exist on servers. In addition, Bill sees other obstacles, such as finding a place to mount an amp as well as power limitations.
We see the most attractive facet of using a ROADM for metro 100G needs as the carriers would not have to buy much, if any, new equipment, which the incumbents would really like, as we noted in our response to a blog post. In the backhaul core, the carrier simply splits off a 100-gig wavelength, and then breaks it down into 10-gigs, for example, either electrically or optically. They will go into an optical cross-connect to be sold off to customers. With the heavily promoted metro system approach, there is no practical way at this time to aggregate a bunch of new 100-gigs in the same way that 10-gigs would be combined into a single 100G.
In the early days, after its initial success in selling DWDM long-haul systems, Ciena was legitimately clueless as to the hardships in penetrating the metro space. Today, despite the rhetoric to the contrary, the vendor lacks any excuse with its vast experience to fail to recognize that a potentially large market for a metro 100G system is not in sight – including at Verizon, which despite promoting an aggressive deployment plan for 100-gig more than any other service provider, it apparently does not even have a nationwide core at that speed, yet.
[written by Mark Lutkowitz]