Metro DWDM

Over the past six months, we've published a series of reports on technologies devised to help telecom operators spruce up their metro networks, so they can offer new and improved services while cutting costs.

So far, these reports have focused on technologies in the electrical domain – Next-Gen Sonet , Metro Ethernet, and Resilient Packet Ring Technology – which leaves the question: What's happening down at the optical layer?

In one respect, a lot is happening. Metro DWDM (dense wavelength-division multiplexing) equipment has come a long way since it was first introduced about five years ago. Vendors have found ways of slashing equipment prices and are now focusing a lot of effort on driving down carriers' operating costs.

In another respect, progress has been disappointing. Metro DWDM still isn't widely deployed in metro networks – although equipment vendors say that interest in the technology is definitely on the increase.

Why the growing interest?

One reason is that metro DWDM is required for the deployment of wave services, which are being provided by a growing number of carriers.

Another reason is that metro DWDM offers a way of solving the bandwidth bottlenecks that are emerging in metro networks as access technologies such as DSL (digital subscriber line) and cable modems are becoming widespread.

A third reason could be that all-optical metro networks still sound pretty sexy – and the first step towards that goal is to get a critical mass of wavelengths deployed.

This report presents the latest developments in metro DWDM technology and reviews whether enough has now been done to get the ball rolling on widescale deployment. It's researched and written by Tim Hills, a freelance technical writer and author of our previous reports on metro technologies.

Here's a hyperlinked summary:

Some of this report digs quite deeply into technological issues. To get the most out of it, you may care to listen in on our archived Webinar on the subject, here.

Here's some background reading that might also prove helpful:

— Introduction by Peter Heywood, Founding Editor, Light Reading
www.lightreading.com— Tim Hills is a freelance technical writer. He may be reached at: [email protected].

Dense Wavelength-Division Multiplexing (DWDM) has been at the cutting edge of optical communications for a long time, as time goes in telecom. In recent years it has transformed the economics of long-distance networks (and wrecked quite a few business plans in the bargain) by underpinning today’s glut in fiber capacity. Unsurprisingly, vendors have leapt to apply the technology’s transforming potential to the metro, and a new generation of metro equipment based on DWDM (and its kid brother, coarse wavelength-division multiplexing – CWDM) has appeared.

Some vendors are getting pretty enthusiastic about DWDM’s potential and reckon that the technology is on the point of becoming a metro bright spot.

“We have seen a real shift in the market dynamics over the last 12 to 18 months. Prior to that, most of our DWDM sales have been direct into enterprises for storage network applications,” says Jack Hunt, director of marketing for storage and photonics at Nortel Networks Corp. (NYSE/Toronto: NT). “Now we are seeing a very strong interest from service providers to really focus on managed wavelength services because they see it as an opportunity to get incremental revenue – it doesn’t really cannibalize their revenues from other lower-speed lines, because a lot of this is storage related and so it is new.”

Hunt points out that big enterprises have been building private networks using C/DWDM over dark or lit fiber for some years, but for the carriers and service providers this has been very much customized, ad hoc work that has taken a lot of time and effort to implement. In business terms it doesn’t scale well, because a single customer has to absorb all the costs. Now, however, metro DWDM technology has progressed to the point where it can support managed wavelength services over a shared infrastructure, which rewrites the cost side of the equation.

Bigger and better

And metro DWDM technology has certainly progressed. Metro wavelength capabilities have moved from a single wavelength to more than 64 wavelengths over the last two or three years; wavelength speeds have increased to 10 Gbit/s; and systems can support rings, meshed rings, and fully meshed architectures. Transmission ranges have increased – for example, through the use of amplification (although the range of some unamplified systems has been considerably increased, too) – so that large DWDM/CWDM rings are possible.

The development of multiservice provisioning platforms (MSPPs) has also become crucially bound up with DWDM, and DWDM MSPPs are now a key product category in the development of metro capabilities and intelligent optics.DWDM MSPPs have several key attributes, including:

Slicker and smarter

But the significance of new-generation metro DWDM goes much further than the potential for all-singing, all-dancing superboxes. The generational shift is from “dumb” DWDM to increased networking functionality at the optical level. As Robin Abel, business manager of photonics at Marconi plc (Nasdaq/London: MONI), points out, the earlier generation of DWDM – say, 24 months ago – was essentially dumb, just providing overlay wavelengths for capacity from one point to another in the metro.

“It doesn’t do anything particularly interesting other than multiplex wavelengths together,” he says. “For certain applications that’s perfectly appropriate, and there is still a space for it. But I think what we are seeing now is the requirement to extend pure connectivity into more networking-type functionality, which may encompass something like switching and routing.”

Such functions reflect increased carrier requirements for smart management capabilities to allow the networking and routing of managed pipes through an optical transmission system.

Switched DWDM starts to make optical networks dynamic and intelligent, and this in turn affects the industry itself. Says Denis Gallant, chief technical officer at Meriton Networks Inc. (formerly Edgeflow), one of the new players in metro DWDM: “We are positioning ourselves as a wavelength networking company, basically taking a network approach to the problem of networking wavelengths in the metro, so we have suite of wavelength networking elements from the metro core to the metro access. And we also have a fairly comprehensive network and service management platform that allows you to support the full suite of element management, network management, and FCAP [fault, configuration, accounting, performance].”

And vendor activity goes beyond new products, technologies and approaches. More metro DWDM equipment is going through Telcordia Technologies Inc.’s Osmine certification in the U.S. This is a crucial step for real metro DWDM, as Osmine (Operations System Modifications for the Integration of Network Elements) governs the interoperability of automated OSS (operations support system) functionality and flowthrough provisioning, and is an essential requirement of the U.S. RBOCs (regional Bell operating companies) and other service providers and carriers.

Last year (2001), two big players and their key products – Cisco Systems Inc.'s (Nasdaq: CSCO) ONS 15454 Metro Optical Transport Platform and Nortel's OPTera 5200 Multiservice Platform – completed the Osmine process for three key systems:

Making waves

But it’s not all technology and whizzy products anymore – the current telecom downturn has made sure of that. Vendors have become much more focused on DWDM’s potential to generate new revenues for carriers and to cut carriers’ costs. Quite a lot of hopes are now riding on the appeal to big corporates of high-bandwidth services using wavelengths and lying outside the bandwidth limitations of the conventional carrier Sonet infrastructure. And adding wavelengths to metro Sonet can lower carrier costs.

“Carriers want to use metro DWDM for basically two reasons,” says Paul Zalloua, VP of marketing for LuxN Inc. "They want to be able to offer high-bandwidth services that they cannot offer on Sonet (Synchronous Optical NETwork) and SDH (Synchronous Digital Hierarchy) networks. In a Sonet/SDH infrastructure you are limited in terms of rate. From a PDH E1/E3 you move to an STM-1/4/16 – there is nothing in between. The second reason is that they want to be able to scale their transport infrastructure. To upgrade Sonet to OC192, you have to upgrade all nodes within that ring. And that becomes very cost prohibitive. But with DWDM you can use the same fiber with multiple channels running.”

According to Nortel’s Hunt, the big attraction of managed-wavelength services for major enterprises is their combination of low latency, high-bandwidth, and private-line characteristics, such as security.“The attraction of wavelength services from a storage perspective is first and foremost the latency and the fact that you can carry Fibre Channel, ESCON, and so on natively,” he says. “The alternative that customers often look at is to downspeed to a lower-speed line, but then you are introducing latency, and for any kind of mission-critical application where latency is important, that’s an issue.”

He cites a case in which a customer downspeeded from the 1-Gbit/s native-mode Fibre Channel to a lower-speed Asynchronous Transfer Mode (ATM) line, and found it reduced the number of transactions processed from 400-500 per second down to 10-20 a second, which was unacceptable.

Despite vendors’ enthusiasm for pushing all the right carrier business buttons, metro DWDM is still generally a limited and patchy market that has yet to see much in the way of sophisticated optical infrastructures. As David Gross, author of the June 2002 report The Market for Optical Bandwidth Services (see CIR Sees Bandwidth Boom) from U.S. market research firm Communications Industry Researchers Inc. (CIR), points out:

“If you look at what has been implemented in the U.S.A. – and not what’s being talked about within the carriers – it’s a lot of point-to-point fiber relief. The optical technology’s now wonderful, but if you start going through on a ring-by-ring basis, there’s very little metro DWDM out there. It’s mainly point-to-point on offices where there is very heavy traffic. And even there they are still stacking a lot of Sonet rings as well.”

In an earlier report, Metro Optical Networking Market Opportunities, published in November 2001, CIR forecast that 2003 would be the first big year for substantial metro DWDM deployment in the U.S., but that the equipment market would not reach a sizeable $500 million until 2005 (see Report Charts Metro Core Growth and CIR Reports on Metro). Major purchasers powering the market will be the four RBOCs – a shift in emphasis, as some of biggest users of metro DWDM to date have tended to be the CLECs (competitive local exchange carriers) building their new networks.

Commenting on the forecast, Mark Lutkowitz, CIR’s VP of optical networking research, said: "It has been the classic chicken-and-the-egg scenario since early 1997, in which the industry has been waiting for a critical mass of wavelengths in the metro space to enable both the optical add/drop multiplexer and optical switching markets to develop. But, until the vendors can provide a combination of lower cost and engineering-friendly DWDM products with true ring capability, the all-optical vision in the metro segment will not begin to be realized."

But metro wavelength services for large enterprises are increasingly available. Pretty well all the main U.S. carriers – such as AT&T Corp. (NYSE: T), SBC Communications Inc., Verizon Communications Inc. (NYSE: VZ), and WorldCom Inc. (OTC: WCOEQ) – offer them, and carriers outside the U.S. are not being left behind. The U.K.’s incumbent carrier, British Telecom (BT) (NYSE: BTY; London: BTA), launched a set of metro wavelength services in June 2002 under the Wavestream banner (see BT Lights Up Storage Service). This comprises:

However, revenues from wavelength services may not be as large as some have hoped. BT estimates it will generate revenues of about $90 million annually by 2004/5 from the new Wavestream services. The experience of the U.S. market suggest that wavelength services will be a niche, although a fast-growing one.

“We forecast domestic metro retail and wholesale wavelength services out to 2006 as a $450 million market,” says CIR’s Gross. “It’s not huge, but it’s a growing market, up from $75 million currently. So there is a lot of growth there, but it’s not a billion-dollar opportunity.”

Gross argues that retail wavelength services for corporates have been oversold, and will be only a small part of the total market for high-bandwidth services. They face the perennial obstacle of a lack of fiber going into buildings. He sees the market limited to a few niche corporate segments and geographical areas – like, for example, brokerage houses in New York and New Jersey

The brighter side of the market is wholesale wavelengths, where constraints on fiber are not an issue and there is considerable potential for dark-fiber replacement. Says Gross: “It’s just a simple cost equation. If you are an ISP and you need to get from a data center to a major NAP that may be 10 miles away, three or four years ago you might have leased dark fiber. Now that’s where wavelengths are making a lot of sense in the wholesale market.”

And, despite metro DWDM’s slowish U.S. start, there is overall growth, especially in developing markets where fiber is thin on the ground and carriers are desperate to increase capacity. A typical recent example is provided by Companhia de Telecomunicacoes do Brasil Central (CTBC Telecom), which is using Lightscape Networks Ltd.'s XDM range of hybrid optical platforms to build a 6,500km DWDM regional network in the east-central region of Brazil. The network will connect major cities, including Brasilia, Sao Paulo, Rio de Janeiro, Belo Horizonte, and Uberlandia, and will have a metro component. Lightscape Networks will also install its small-capacity XDM-1000 and XDM-500 platforms in several of the metro networks to support the future expansion of SDH/Sonet there (see Brazil Picks Lightscape's XDM).

The original idea behind DWDM was the simple one of increasing fiber capacity by increasing the number of optical channels each fiber could carry. On the basis of one optical channel per wavelength, this is done by multiplexing several wavelengths onto a fiber. A number of metro vendors, particularly from the Sonet side of the industry, use DWDM essentially as a straightforward add-on to create multiple parallel networks, where DWDM’s role is largely passive.

As Dana Hartgraves, VP for product marketing at Metro-Optix Inc., explains: “We play into the DWDM world by offering the ITU Grid optics at the OC48 rate – and will do so also when we have OC192. We offer DWDM as an OEM device in a separate one-increment shelf. It is as a passive DWDM system, but we can always play into an active switched DWDM system as well.”

An obvious next step is to treat some of the additional optical channels as transparent (protocol-independent) transport capacity and to add new service capabilities – such as native-mode high-speed data (in particular Ethernet and the storage protocols Escon, Ficon, and Fibre Channel) – to them. This allows carriers to support their current Sonet/SDH infrastructures while introducing new services in a pretty flexible fashion, being transparent to protocol technologies, and has obvious attractions in the protocol-rich metro environment. But this will be most effective only when carriers can switch the transparent capacity. Which takes DWDM on to another stage of capabilities.

We are entering an arms race here, as DWDM quickly acquires a range of increasingly sophisticated (and ambitious) ideas. Once we are talking of switching optical channels, it becomes very tempting to integrate functions optically so as to minimize, say, the number of ports or OEO (optical-electronic-optical) conversions in an effort to reduce cost, footprint, and power consumption. Having done this, hiding the optics from the carrier and adding optical-management/service-intelligence to simplify operations becomes pretty well essential if carriers are to cope with the complexity while reducing costs.

Tangle of terms

Before getting carried away with the possibilities, it’s best to backtrack a little and clarify some terms, as metro C/DWDM comes in different flavors, depending on which vendor you are dealing with.

Wavelength Division Multiplexing (WDM) comes in two main forms in the metro: dense (DWDM) and coarse (CWDM), the difference being that DWDM packs far more optical channels into the fiber transmission window than does CWDM. Although this report is mainly about DWDM, CWDM looks set to have a growing role within practical DWDM network architectures. So it’s a good idea to check out the CWDM aspects of any DWDM system – and whether CWDM could not do the whole job on its own, anyway.

Both DWDM and CWDM are, in principle, protocol transparent because the basic technology is concerned with creating and managing optical wavelengths or channels and does not interpret the transmitted optical bitstream at a higher level. However, in many practical systems there is a complication that slightly upsets this simple picture – digital wrappers. So it is necessary to distinguish between wrapper and wrapperless systems.

The point of a wrapper is to provide a framed optical channel on a wavelength. This framing can be used for three main things:

As its name suggests, a digital wrapper adds extra framing bits into the optical signal, effectively wrapping the client signal into a digital envelope, but otherwise not disturbing it, so as to maintain transparency. Digital wrappers tend to be proprietary to the vendors concerned, but standards for DWDM/all-optical networks and digital wrappers are now falling into place. The International Telecommunication Union, Standardization Sector (ITU-T) approved a number of Optical Transport Network (OTN) standards and revisions to standards during 2001, including G.709 and G.959.1, which address the interface frame format and physical layer interfaces, respectively. G.709 preserves many of the ideas of Sonet/SDH, such as in-service performance monitoring, protection, and other aspects of management. Additional functions include optical wavelength management end-to-end – obviating the need for intermediate electro-optical conversions – and FEC to enable longer optical spans.

So the result of optical-framing/wrapping in DWDM systems is to produce a wavelength-aware framing scheme that can act as the managed base transport layer for all the other major lower-layer protocols – especially for Sonet/SDH, Gigabit Ethernet, ATM, Fibre Channel, and Generic Framing Procedure. However, not all vendors are convinced that digital wrappers are always necessary for every type of metro application – there is inevitably a processing overhead, for example – and a number of current DWDM systems are wrapperless. They provide some of the wrapper’s functions, such as channel management, by other means. (For more details about digital wrappers, see our Beginner's Guide: Digital Wrappers and Forward Error Correction (FEC).)

Single or multiprotocol

Another distinction between systems concerns the two ways of exploiting wavelengths for transport at the system client-signal level – single protocol and multiprotocol. Single-protocol systems support only one protocol per wavelength, although they are also multiprotocol in the sense that different protocols may be running on different wavelengths simultaneously. Close reading of the system specification is sometimes needed to distinguish this meaning of multiprotocol from the more functional one where a single wavelength can support different protocols simultaneously.

A practical point with systems that support multiple protocols on a single wavelength is that you can do clever things in wavelength traffic management (grooming and loading efficiency and so on) but at the cost of a more complex system. And this type of multiprotocol capability may be unnecessary if your main interest is just providing clear-channel wavelength services.

Single-protocol systems are also often marketed as providing wavelength gain, which is another way of saying that that they can offer either a higher level of multiplexing than occurs within the client signal’s normal multiplexing hierarchy or a different type of multiplexing step. Sweden's Lumentis AB, for example, has recently released a one-slot dual GigE transponder that aggregates two Gigabit Ethernet signals onto a single wavelength; and the company’s Muxponder aggregates eight STM1s and two STM4s onto a wavelength.

Finally, it’s worth noting that managed wavelengths and wavelength switching can be interpreted in several ways. The most basic ability is to provide an optical path through a network that uses the same wavelength on every link traversed. This amounts to the ability to select one wavelength out of the set supported and to connect two endpoints across the network by switching that wavelength through the appropriate links. A big problem with this approach is that it gets increasingly difficult to do as the number of customers increases; it lacks flexibility; and it can lead to wavelengths being stranded, which is highly inefficient.

More useful, and more advanced, is the ability to used different wavelengths on different links to support the optical path, so the switching here divorces the optical channel from the wavelengths supporting it. This is the usual mode of operation. Being able to map channels to wavelengths also has cost benefits – for example, three or four wavelengths from different access points can be mapped into a common wavelength band across the metro core while jointly applying amplification, which is cheaper than amplifying each wavelength separately.

Wavelength switching of this type can be done either entirely optically (known as optical-optical-optical or OOO) or through an intermediate electrical stage (OEO). OEO systems are currently more common, largely for reasons of cost, although the two-stage OE and EO signal conversion increases complexity and lacks the elegance of the purely optical process, which is more in accord with the idea of an optical transmission layer. Note that bit-level OEO is protocol transparent, as is OOO.

Wrapper systems can complicate the switching terminology even further, because some can identify individual client signals within a multiprotocol stream on one wavelength and can in principle switch them at the client level into another stream on a different wavelength. While this is not the same as switching at the optical-channel or wavelength levels, it gives a similar effect as far as the client applications are concerned.

Coarse wavelength-division multiplexing uses a greater spacing between wavelengths (typically 20 nanometers) than DWDM does (typically 0.8 nm). This means that fewer channels can be accommodated in the optical transmission band, so the overall system transmission capacity is lower – but the optics can be much cheaper as it works to much larger tolerances. For example, lasers do not need cooling to maintain wavelength stability, and this can save thousands of dollars in the cost of the optical subsystems. Straightforward transceivers using standard Distributed Feedback (DFB) Lasers can be used instead of the discrete lasers and custom-built electronics typical of DWDM systems. Lower power consumptions make board layouts simpler, and in, general, the whole production process is easier and therefore cheaper, too. Typically, CWDM can be about one third the channel cost of DWDM.

This fact in today’s cash-strapped environment is one of several that is leading to greater interest in CWDM, which in the past has been seen as something of an orphan among metro technologies and only of limited interest. CWDM is now emerging from being a specialist niche to having much wider applications, and the technology is being improved and standardized.

Lars Bergström, segment marketing manager of Transmode Systems AB says, “The new ITU-T standard is proof that CWDM is a mature technology. It’s no longer an exotic new technology. It is good proof to the market that CWDM is here to stay, and the word is out in the market about CWDM. It wasn’t long ago that no one knew anything about CWDM, and we spent all our time trying to educate people.”

Major developments include:

As a result, metro CWDM is becoming potentially important for a wider range of players, including:

Metro access is likely to be one of the most important metro applications of CWDM. Many of the major vendors use CWDM as the access portion of their metro DWDM systems because lower costs are usually far more important than capacity in this part of the network. So growing numbers of vendors now offer CWDM as an option or additional component within their metro DWDM systems. This can lead to various hybrid schemes – LuxN, for example, offers a facility to combine 16 C-band DWDM wavelengths with four non-C-band CWDM wavelengths for a total of 20 wavelengths. The company points out that it can also crossconnect CWDM access rings into DWDM core rings, providing both low cost for the access network and high scaleability for the metro core network, with end-to-end management for each wavelength.

Internet Photonics Inc. is one company following a combined C/DWDM approach, its MXA access system using CWDM; its core MX, DWDM.

“We view CWDM as the typical way that you would have the access to the metro network edge. We would use DWDM more typically in the IOF between the CO and the core networks,” says Gary Southwell, Internet Photonics’ VP for marketing. “The power of CWDM is that we can further drive the cost point down. It gives us a good cost point, security, the ability to put, in our case, up to eight wavelengths in the access loop. We find we have more than enough capacity for multiple customers, as the typical access loops that are out there today only have one, maybe two, customers on them. As they grow over time, we would consider going to more of a DWDM solution by using our MX platform.”

The introduction of amplified systems, such as Transmode’s add-on Linear Optical Amplifier (LOA), is literally extending the range of applications by extending the reach of the system. “We are still not trying to do long haul – that’s not the point of it – but we are targeting regional networks,” says Bergström. “One of our customers has a pan-Scandinavian network, so we have installed our system with fairly long spans, but we have had to use repeaters. The LOA would now be a much more cost-effective way of doing it.”

Despite the rising interest in CWDM, the industry is concentrating much of its efforts on DWDM. And there is a lot to do, as DWDM equipment is inherently more complicated than conventional single-wavelength metro systems, and it introduces an optical dimension into the end-customer relationship. Key capabilities and features for carriers that raise issues include:

While DWDM optical capabilities continue to develop, for current metro applications today’s optical technology is generally pretty good. Pushing forward the leading-edge performance in, say, wavelength packing densities is unlikely to be a priority any time soon. Greater integration, along with smaller footprints and lower cost and power consumption, will be welcome – but much of the focus is now at the system-level functions.

As LuxN’s Zalloua observes: “Six months ago I would have said carriers were interested in more than 64-wavelength DWDM, but right now nobody is talking beyond 64 wavelengths – nobody is talking beyond 32 in the metro. Thirty-two is the comfort zone right now. Anything above that is good, but they are not going to use it in the near future.”

But obviously, development continues, especially at the card level. One recent step is to offer frequency-agile lasers that are tunable across the entire C-band, as Lightscape Networks has just done by integrating Agility Communications Inc.’s 4245 EML widely-tunable laser into its XDMT hybrid optical networking platform. This, the company says, gives a tunable line-card that supports 50GHz and 100GHz channel spacing and enables dynamic configuration and reconfiguration of wavelengths while in service, without affecting traffic on the network (see Clouds Lift on Tunable Lasers).

"Cost reduction is a major concern for today's vendors and their customers," says Ido Gur, VP of marketing at Lightscape. “This development gives operators the opportunity of massive economies of scale across the network, as a few tunable lasers replace the need to hold large numbers of expensive spare fixed lasers. This simplifies inventory management and ultimately saves on truckrolls. Failed transmitters can be easily restored and wavelengths rerouted, eliminating the time delay normally associated with installing additional cards for the routing and rerouting of traffic.”

Because the ramifications of metro DWDM are so large, vendors are unsurprisingly emphasizing different aspects as a means of product differentiation. This naturally affects both the general type of system being offered and the details of the particular technologies used. Of course, certain themes – such as improved management – are emphasized by pretty much all vendors.