Optical IP Revisited

It’s been interesting following the discussion of FTTP over the past 18 months. From questioning whether telecom investment is being driven by politics or competitive zeal to debating whether RBOCs will move to GPON or EPON, the focus has all been on the access network.

At OFC this year, every session that had a panelist from an RBOC was barraged with questions on PONs. The anxiousness of it all revealed just how many device vendors’ futures will be tipped up or down by the answers.

Are we missing the point? The impact on the larger network seems to be more important. But ask these folks about what impact triple play and FTTP will have on the network behind the central office, and most have a fuzzy answer at best –- more Gigabit Ethernet, more WDM, or, simply, that it’s under study today.

I’ve had time to think a lot about this. In putting together the program for this year’s “Future of Optical Networking” live event, it’s become clear that a lot of this activity –- and a lot of the impact of FTTP –- will be focused in metropolitan networks. Certainly PONs have gotten a great deal of attention in the optical networks market this past year. But in the long run, optical access networks must be viewed in light of their impact further downstream, on the entire network.

Optical networking has long promised a highly flexible, switched, and intelligent network, but the bulk of its delivery so far has come simply in the form of cheaper pipes, leaving the real networking to electrical systems in Layers 2 and 4. The grandiose visions of a switched photonic network have largely remained just that -- grandiose visions trapped in trials and testbeds, with a few implementations that you would need Defense Department clearance to see. The problem, in hindsight, was relatively simple: If services hadn’t fundamentally changed, then why should the transport network?

But this year, things are looking a little different. Triple-play networks, deployed by both telcos and cable MSOs, are in fact fundamentally different from their predecessors. These networks will ultimately carry broadband data, packetized voice, broadcast video, and interactive video services, all over an IP network. What was once a network based on a series of point-to-point relationships between customer and various local endpoints (Class 5 voice switch, T1 mux, DCS, etc.), will become much more distributed, multipoint, and broader in scope.

The introduction of video on demand alone requires significant changes in the way metro networks are designed. For one, these networks will need significantly more bandwidth, and it needs to be cheap. It requires a mixture of Ethernet and WDM. Many hardware companies have been heading in this direction. At the “Future of Optical” event we’ll be hearing from Corrigent Systems Inc., which has a product that marries RPR, MPLS, pseudowires, and 10G Sonet to create a metro aggregation and backhaul network that can handle a variety of services in a unified fashion, with multicast efficiencies. It’s selling the product to KDDI Corp. Fujitsu Network Communications Inc. (FNC) has had its successes with ROADM systems in the MSO market and is now looking to take on telco triple-play networks, arguing that the metro core will require optical-layer reconfigurability. We’re hearing similar stories from Mahi Networks Inc. and Meriton Networks Inc. Plus, Lambda Optical Systems Corp. and Infinera Inc. are sending some rather daring messages -- both proposing very different, yet both rather radical, alternative to traditional networking solutions. CoreOptics Inc. and Intel Corp. (Nasdaq: INTC) will be on hand as well, showing the way towards a new generation of transponders that takes advantage of advances in tunability, integration, and electronic dispersion compensation.

The overwhelming impression is that the optical market is moving cautiously forward, from a period of focusing exclusively on cost reduction to new levels of intelligent optical layer networking, at least for systems in the metro core and regional networks.

As the following chart shows, there is some support for this notion from service providers. For an upcoming Heavy Reading report on reconfigurable optical networks, I surveyed over 60 service provider representatives and talked to a dozen directly.

The results were somewhat of mixed bag and the timing of this investment is not yet clear, but there is an indication that a transformation to the intelligent switched metro optical network is inevitable. Here are a few highlights I had to mull over before finalizing our key findings:

  • The network is growing, period. Operators reported growth from 10 to 25 percent across their entire transport networks. They cited business Ethernet services and triple play as the key drivers.

  • The service providers are not in agreement on the best way to accommodate this growth. Solutions included integrating ITU-grid optics into Sonet devices, adding Sonet aggregation switches to the metro core, adopting simple WDM solutions, and adding more sophisticated ROADM based solutions. This is good news for vendors with lots to offer, but not so good news for vendors of point solutions.

  • The major RBOCs in North America definitely want to move forward with “ROADM Systems” and have RFIs and RFPs out there, with spending likely to be around $25 million to $30 million each per year as they build out their metro cores. Their wish list of features is a mile long, yet they don’t appear very excited to pay the premium. In the meantime, vendors of traditional WDM gear are doing just fine and don’t feel a real urgent need to jump into the ROADM game just yet.

  • Tier 2 and 3 operators committed to ROADM systems are hard to find. Most remain focused on low-cost metro solutions and haven’t seen a need to move to ROADMs just yet. They like the idea of ROADM, but that’s about it.

  • Those operators who like ROADMs like them for their opex values, ranking one-time network or node engineering and simplified wavelength provisioning as key values. Problem is, selling systems based on their opex value has always been tough.
  • It’s hard to find much enthusiasm for ROADM systems outside of the U.S. Most European operators (save BT) continue to focus on metro WDM as a solution for high-bandwidth services to large enterprise customers, while in Asia there are a few operators building some interesting switched optical networks, but in general, it’s hard to find a similar crop of ROADM boosters as you will here.

  • But, before calling out ROADMs as just another hyped optical technology, it is worth considering that even though demand is very fragmented today, nearly all operators were unanimous in saying that the current crop of WDM systems out there is a pain and will not be used to build new triple-play optimized metro networks until these limitations are addressed. You get the feeling that ROADM systems are in many ways inevitable, and will ultimately displace traditional metro DWDM in the metro core, but carrier sensitivity to price premiums will keep this space slowly rolling for a couple years. Probably sounds familiar to anyone selling tunable lasers over the past four years.

So, what does this all add up to? Let’s call it the “IP Services Network,” which we’ll be talking about at the event next week. This is a two-layer network vision: A packet services layer carries all services as IP, and beneath that lies an optical layer with a degree of automated wavelength-level management and network intelligence to support the service layer’s requirements for reach, protection options, and scaleability into the hundreds of gigabits. Oh, and that's the part that vendors hate to hear: It should cost less than today’s gear. — Scott Clavenna, Chief Analyst, Heavy Reading

Sign In