Inside Telefonica’s Optical Announcement

Telefonica today announced a research-network trial of a basic IP/DWDM technology combination that flattens networks to an agile IP/MPLS edge and an agile optical core built around GMPLS.  The project included Universal Edge routers from Juniper and Adva Optical DWDM core technology.  Everyone is happy about the resulting reduction in capex and opex, and in the speed of creating new services, but there’s more to this than the press releases talk about.

As the Internet has driven down revenue per bit, operators have struggled to compensate by reducing infrastructure cost per bit to stabilize ROI.  Traditional router networks, which are a hierarchy of edge and core devices, increase capital cost and operations cost and they don’t really provide capacity, they provide connectivity.  If agile optics could be used to create a flexible core network that could respond to failures, the same technology that actually creates bits could connect the edges.  This has been behind a wave of operator RFIs on optical networking, notably Verizon’s OTN initiatives, sometimes as much as five years old.  The essential concept behind agile optics is that if you could create a wavelength mesh of edge devices there’d be no need for core connectivity management.

From the perspective of the core network builder and the user, this is a good thing because it improves ROI on infrastructure.  But there are some limits to the goodness, and the most obvious is for the network equipment vendor.  What tends to happen in OTN transformations is that the IP devices become little more than big on-ramps.  That means the total investment in core infrastructure shifts more toward optics, whose overall portion of the budget is likely to increase, and away from IP.  IP has higher margins, so that’s bad for vendors, particularly those who don’t play in the optical space at all.

A second part of the picture is highlighted by that comment about the MSAs.  Today, most Internet traffic is shortstopped within an MSA because it’s either hitting a local resource or it’s hitting a content cache.  Video traffic is almost never hauled over the core, and that’s where most traffic growth is concentrated.  Metro networks are different from core networks in that they’re really AGGREGATION NETWORKS and not connectivity networks; their focus is to concentrate traffic toward points of service presence.  That’s never been an IP mission, it’s typically more about Ethernet and tunnels.  So traffic growth overall is shifting emphasis from IP connectivity to Level 2 aggregation, and that also diminishes the role of routers.

Then there’s the critical point.  We’ve got a market driven by a different kind of aggregation; the summarization of billions of small, low-value, relationships into a gigantic traffic pit we call “the Internet”.  It’s so big that planning for it consumes all planning cycles, and so pervasive that its economy of scale is unbeatable and everyone designs apps for it rather than for other services with better service but higher cost.  In this market, every vendor has to accept that the network is going to dumb down, generate lower cost per bit, become more optical and un-differentiable.  They have to decide what their revenue and profit sources will be in the new age.  Are there models, things we’d call “SDNs” today that could reignite network value?  Sure, with proper foundation in regulation and business practices and with proper support for cloud-hosted features.  We need to transition all of the network to the optical age, not just the core.

A final point here.  If you like this kind of blog coverage of announcements and issues, we’re going to be doing more stuff you’ll like.  We’re working on a video-based (Google+) framework for delivering commentary on what we believe to be key announcements and issues.  The current plan is that we’ll set a specific regular time (provisionally 11 AM Eastern Time on Friday) and then set up a Hangout.  In some, I’m going to be the only speaker.  In others we may have journalists, vendors, or even other analysts.  If you’re interested in this sort of thing and want to be kept advised, drop an email to hangmeout@cimicorp.com and we’ll keep you posted.

Leave a Reply