Voodoo Econ in Metro DWDM

There is no doubt that metro-area networking is still hot, but what has happened to the craze for metro-area DWDM? Well, as it turns out, an important shift may be at hand.

Five years ago, every pitch I saw on metro DWDM dealt with the delivery of wavelength and Gigabit Ethernet services. Over the last two years, however, the pitch has shifted to cost savings. Funny how the marketing pitch shifts once the penny pinching begins.

I have lost count of the number of business cases I have seen touting the tremendous savings of DWDM over Sonet ADMs. With the tremendous push to cut costs in carrier networks, one would think there would have already been a mass migration to DWDM. So why, other than specific cases of point-to-point fiber relief, is Metro DWDM failing to make the grade?

Well, it turns out that business-case theory is one thing and practical reality is another. There are two areas where these scenarios diverge: The first lies in the difference between the first cost of equipment versus long-term cost, and the second concerns the operational impact that comes from the inflexibility of metro DWDM versus Sonet.

“The vendors usually show a business case that compares fully or substantially loaded systems,” says Sam Greenholtz of Telecom Pragmatics Inc. Greenholtz, now a telecommunications consultant, was involved in the initial rollout of DWDM systems used for fiber relief at Verizon Communications Inc. (NYSE: VZ). “The savings may not come until years down the road. The RBOCs are just as concerned with first cost, especially in these economic times. The vast majority of DWDM deployments by RBOCs are for niche applications for fiber relief.”

The second issue is the inflexibility of DWDM rings when compared with Sonet. Most DWDM optical add/drop multiplexer (OADM) systems add and drop wavelengths in fixed bands (or groups), typically four at a time. Banding not only contributes to the high initial cost, but also requires complex planning to make sure that wavelengths are used efficiently. In addition, most DWDM systems provide little if any management data on wavelengths that are not dropped (i.e., that are passed through) as compared to that typical of Sonet systems. This can make it hard to trace problems with individual channels.

This inflexibility is particularly unforgiving of an error in traffic forecasts. “When a forecast is missed for a wavelength, an overlay network is needed to serve the change in forecasted demand,” says Scott Mountford, Corporate Manager of Fundamental Network Planning at SBC Communications Inc. (NYSE: SBC). “The overlay network can cost in the millions of dollars.” These factors are never considered in the theoretical DWDM business case presented by most vendors.

In short, the theoretical business case has been based on quite a bit of smoke and mirrors. That’s the bad news. The good news? The vendors are listening and are beginning to develop systems based on Reconfigurable OADM technology. These systems allow wavelengths to be added and dropped individually on a dynamic basis rather than in fixed groups. This dramatically cuts both the capital and operational costs. “ROADM allows the flexibility of Sonet-like management of wavelengths,” says SBC’s Mountford.

Basic ROADMs are only a part of the answer, however. ROADM component technology has been around for a while, but ROADMs introduce an even stronger need for management data on individual wavelengths. ROADM rings are very sensitive to topology changes and require close monitoring and control of wavelength power to keep the system in balance. In addition, some sub-rate multiplexing is needed to efficiently utilize wavelengths and so strengthen the business case. These issues are being addressed, however, and the business case can be compelling. Mountford says that ROADM technology combined with integrated Sonet multiplexing for optical input easily beats Sonet for metro-to-metro interconnects, and his modeling shows that it also proves in for three to six node rings within the metro.

Several RFPs for ROADM-based solutions are expected within the next three to six months, and most major vendors have begun the development of non-banded ROADM-based systems. Startups Tropic Networks Inc. and Photuris Inc. are focused on next-generation ROADM-based systems and appear to have a substantial jump on the larger vendors.

Will this result in any business, or are these just more lab experiments and RFP exercises? If some of the service provider models are correct, ROADM-based systems not only move the business case from negative to positive, but also may save hundreds of millions of dollars over the next few years. These kinds of numbers are generating a lot of interest at the executive level, not just in the labs.

This time, the business case numbers generated by the vendors and the RBOCs just might match. It should be interesting to watch.

— Doug Green is founder and principal of the Bradam Group LLC, a telecom consultancy in northern Virginia. He can be reached at: [email protected].

Many of the issues discussed above are covered in greater depth in the inaugural report from Light Reading's independent market research division, Heavy Reading: Multiservice Provisioning Platforms: Empowering the Metro Edge.

Page 1 / 2   >   >>
vrparente 12/4/2012 | 11:27:40 PM
re: Voodoo Econ in Metro DWDM One revision. I should never say "I." It was more like "my team and our partner vendors."

vrparente 12/4/2012 | 11:27:40 PM
re: Voodoo Econ in Metro DWDM I built two metro DWDM systems initially 16 ch. each (later upgraded to 32 channels). In the first two years these systems payed for themselves in cost savings over leased fiber and leased SONET (the alternative). That was 12 months ahead of 36 month capital depreciation schedule which was already significantly more advanced than any schedule on leased SONET equipment. And add to that the value of the systems being 100% up for over 24 months -- beating leased services by nearly 0.1%.

But I will add that these systems are somewhat less flexible. That is, they were used to provide transport services that were infreqently changed.
gea 12/4/2012 | 11:27:38 PM
re: Voodoo Econ in Metro DWDM I didn't see this explicitly mentioned, and it is probably as important as the reconfigurability issue. SONET systems can protect within approximately 10ms, in standardized BLSR, UPSR and linear configurations. Amplified DWDM systems can do this, but it's VERY hard, for reasons related to the ROADM issues (BTW, this is what ONI did and because of that it was fairly unique.)

The interesting thing will be to see CWDM systems with protection capabilities emerge...no EDFAs and standardized interfaces.
vrparente 12/4/2012 | 11:27:34 PM
re: Voodoo Econ in Metro DWDM Adding to my previous post and reply to the last. We did have optical protection switching on the systems we deployed. In regional (300km size) rings, it worked sufficiently fast (<60ms).
gea 12/4/2012 | 11:27:33 PM
re: Voodoo Econ in Metro DWDM "We did have optical protection switching on the systems we deployed. In regional (300km size) rings, it worked sufficiently fast (<60ms)."

Actually, that's within the upper limit of SONET PS times, so I agree that's sufficiently fast.

However, the million dollar question is, what was the protection architecture? Was it 1+1, or 'O-BLSR'? And if 'O-BLSR', then the conditions under which that will be cheaper than SONET will be exceedingly rare.
garibaldi 12/4/2012 | 11:27:31 PM
re: Voodoo Econ in Metro DWDM O-SNCP1+1 protection scheme termed 'FlexRing' - a fiber protection scheme for rings with switching time 5-20 ms.
douggreen 12/4/2012 | 11:27:30 PM
re: Voodoo Econ in Metro DWDM Gea,

The problem you bring up is valid, but it is not nearly so hard to solve as the technical challenges to ROADM. When you allow any wavelength to be dropped at any point in the network, wavelength power management (in an economic form) is a huge challenge.
douggreen 12/4/2012 | 11:27:30 PM
re: Voodoo Econ in Metro DWDM Victor,

Thanks for the insights. Note that I am not claiming that metro-DWDM has never proved in. In fact, I helped sell it into both Verizon and BellSouth when I was at Ciena. BellSouth said that it was the smoothest deployment of new technology they had ever seen in their transmission network. The issue was that it proved in for for point to point fiber exhaust. It was used in "special cases" and still is.

Your scenario would also probably be considered nusual in the eyes of the RBOCs.

The hard case comes when an RBOC or MSO, who owns their own fiber, tries to make a more generalized case for mass deployment. This is what is needed to make a market big enough to support, well, a real market.

Nothing is certain in these times, but IMO the flexibility and cost structure of the new systems have a good shot at moving metro DWDM from special case to main stream. If I am right, a couple of startups have caught the big guys with their pants down.
vrparente 12/4/2012 | 11:27:29 PM
re: Voodoo Econ in Metro DWDM Doug,

Agreed. For the most part these systems were "engineered" ahead of time and had low rates of change (mostly adds). And planning wavelength use is a bit tricky. But compared to someother architectures with band drop (like the Cambrian/Nortel, et. al. -- the Ericsson system was much simpler to engineer). And I also agree about the power management. I think there are two practical ways of dealing with existing hardware and fiber. One is good contracts for fiber. The other is keeping the number of nodes down and the circumference down. And I think that means -- as you have implied it is only practical for high density core type functions, which is exactly what we've seen. "Internal" use rather than for customer circuits. There are lots of BIG customers in major metro areas though that have that need for a few nodes with high data rate circuits (OC-12 to 10Gig) that would fit perfect.

We (AOL) had planned a second project (ONI/Ciena) that was later deployed after I left. And the third was the Qeyton/Cisco system. Now that one would have been more "access circuit" type architecture, but I exited before that system was available to purchase. Although we did field test the Qeyton (now Cisco) system and it seemed quite flexible.

If we look at what are now called MCAS systems -- I think we all agree it would be nice to see something like that with optical. But after all, that is like a third/fourth/fifth (depends on how you count I guess) SONET system. And optical is clearly behind that.

And to garibaldi, you either know me or what I was talking about because it was indeed Ericsson FlexBus/Flexring. So who is garibaldi ?

-Victor Parente-Blake

I'm glad to hear Abovenet (MFN) coming out of the file (aka chapter x). They were using LuxN gear to provide customer optical transport. Of course they had also deployed the Nortel/Cambrian solution as far back as late 1999/early 2000 and I think the band dropping architecture made it hard. But after all its just like trunking STS's.

PO 12/4/2012 | 11:27:22 PM
re: Voodoo Econ in Metro DWDM douggreen: The problem you bring up is valid, but it is not nearly so hard to solve as the technical challenges to ROADM. When you allow any wavelength to be dropped at any point in the network, wavelength power management (in an economic form) is a huge challenge.

PO: The complexities of steady-state power management pales in comparison to the complexities of power management in the face of various network failure events, which introduce spikes and a new (difficult to predict) steady-state. After all, nobody wants a single failure to take out unrelated services.
Page 1 / 2   >   >>
Sign In