Lumentum ships a 400G CFP2-DCO coherent module
Lumentum has started supplying customers with its CFP2-DCO coherent optical module. Operators use the pluggable to add an optical transport capability to equipment.
The company describes the CFP2-DCO as a workhorse; a multi-purpose pluggable for interface requirements ranging from connecting equipment in separate data centres to long-haul optical transmission. The module works at 100-, 200-, 300- and 400-gigabit line rates.
The pluggable also complies with the OpenROADM multi-source agreement. It thus supports the open Forward Error Correction (oFEC) standard, enabling interoperability with oFEC-compliant coherent modules from other vendors.
“We are encountering a fundamental limit set by mother nature around spectral efficiency,”
“Optical communications is getting more diverse and dynamic with the inclusion of the internet content providers (ICPs) alongside traditional telecom operators,” says Brandon Collings, CTO at Lumentum.
The CFP2-DCO module is being adopted by traditional network equipment makers and by the ICPs who favour more open networking.
CFP2-DCOs modules from vendors support the OIF’s 400ZR standard that links switching and routing equipment in data centres up to 120km apart and more demanding custom optical transmission performance requirements, referred to as ZR+.
So what differentiates Lumentum’s CFP2-DCO from other coherent module makers?
Kevin Affolter, Lumentum’s vice president, strategic marketing for transmission, highlights the company’s experience in making coherent modules using the CFP form factor. Lumentum also makes the indium phosphide optical components used for its modules.
“We are by far the leading vendor of CFP2-ACO modules and that will go on for several years yet,” says Affolter.
Unlike the CFP2-DCO that integrates the optics and the digital signal processor (DSP), the earlier generation CFP2-ACO module includes optics only, with the coherent DSP residing on the line card.
The company also offers a 200-gigabit CFP2-DCO that has been shipping for over 18 months.
As a multi-purpose design, Affolter says some customers want to use the CFP2-DCO primarily at 200 gigabits for its long-haul reach while others want the improved performance of the proprietary 400-gigabit mode and its support of Ethernet and OTN clients.
“Each of the [merchant] DSPs has subtly different features,” says Affolter. “Some of those features are important to protect applications, especially for some of the hyperscalers’ applications.”
Higher baud rates
Lumentum did not make any announcements at the recent OFC virtual conference and show regarding indium phosphide-based coherent components operating at the next symbol rate of 128 gigabaud (GBd). But Collings says work continues in its lab: “This is a direction we are all headed.”
The latest coherent optical components operate at 100GBd, making possible 800-gigabit-per-wavelength transmissions. Moving to a 128GBd symbol rate enables a greater reach for the given transmission speed as well as the prospect of 1.2+ terabit wavelengths.
This means fewer coherent modules are needed to send a given traffic capacity, saving costs. But moving to a higher baud rate does not improve overall spectral density since a higher baud rate signal requires a wider channel.
“We are encountering a fundamental limit set by mother nature around spectral efficiency,” says Collings.
Optical transmission technology continues to follow the familiar formula where the more challenging high-end, high-performance coherent systems start as a line-card technology and then, as it matures, transitions to a more compact pluggable format. This trend will continue, says Collings.
The industry goal remains to scale capacity and reduce the dollars-per-bit cost and that applies to high-end line cards and pluggables. This will be achieved using greater integration and increasing the current baud rate.
“Getting capacity up, driving dollars-per-bit down is now what the game is going to be about for a while,” says Collings.
Whether the industry will go significantly above 128GBd such as 256GBd remains to be seen as this is seen as a technically highly challenging task.
However, the industry continues to demand higher network capacity and lower cost-per-bit. So Collings sees a couple of possible approaches to continue satisfying this demand.
The first is to keep driving down the cost of the 128GBd generations of transceivers, satisfying lower cost-per-bit and expanding capacity by using more and more transceivers.
The second approach is to develop transceivers that integrate multiple optical carriers into a single ‘channel’. A channel here refers to a unit of optical spectrum managed through the ROADM network. This would increase capacity per transceiver and lower the cost-per-bit.
“Both approaches are technical and implementation challenges and it remains to be seen which, or both, will be realised across the industry,” says Collings.
100-gigabit PAM-4 directly modulated laser
At OFC Lumentum announced that its 100-gigabit PAM-4 directly modulated laser (DML), which is being used for 500m applications, now supports the 2km-reach FR single-channel and FR4 four-channel client-side module standards.
This is a normal progression of client-side modules for the data centre where the higher performance externally-modulated laser (EML) for a datacom transceiver is the one paving the way. As the technology matures, the EML is replaced by a DML which is cheaper and has simpler drive and control circuitry.
“We started this [trait] with the -LR4 which was dominated by EMLs,” says Mike Staskus, vice president, product line management, datacom at Lumentum. “The fundamental cost savings of a DML is its smaller chip size, more chips per wafer, and fewer processes, fewer regrowths.”
The company is working on a 200-gigabit EML and a next-generation 100-gigabit DML that promises to be lower cost and possibly uncooled.
Reconfigurable optical add-drop multiplexers (ROADMs)
Lumentum is working to expand its wavelength-selective switches (WSSes) to support the extended C-band, and C- and L-band options as a way to increase transmission capacity.
“We are expanding the overall ROADM portfolio to accommodate extended C-band and more efficient C-band and L-band opportunities to continue to build capacity into ROADM networks,” says Collings. “As spectral efficiency saturation sets in, we are going to need more amplified bandwidth and more fibres, and the C- and L-bands will double fibre capacity.”
The work includes colourless and directionless; colourless, directionless and contentionless, and higher-degree ROADM designs.
Has the restructuring of the optical industry already started?
The view that consolidation in the optical networking industry is needed is not new. For a decade, ever since the end of the optical boom in 2001, consolidation has been called for and has been expected. And while the many optical startups funded then have long exited or been acquired, the optical industry continues to support numerous optical networking and component generalist and specialists.
Given the state of the telecom market, is a more fundamental industry restructuring finally on its way?
"The business model of the communication sector needs to change, and change in a relatively short order"
Larry Schwerin, CEO of Capella Intelligent Subsystems
Larry Schwerin, CEO of Capella Intelligent Subsystems, believes change is inevitable. He argues that the industry supply chain will change, especially as firms become more vertically integrated.
"This is not to say that the market and demand are not there," says Schwerin, but the industry is stuck with a decade-old structure yet the market has changed.
Optical market dynamics
Schwerin starts his argument by highlighting certain fundamental drivers. IP traffic continues to grow at over 30% a year, while the nature of the traffic is changing, especially with cloud computing and as users generate more digital media content.
“The current rate of bandwidth growth coupled with the rate of CapEx spend, the gap is widening and the revenue-per-bit is dropping,” he says. “Some argue that bandwidth growth will slow down as operators charge [users] more, but to date this hasn't been seen.”
These trends are welcome for the optical companies, says Schwerin, as operators adopt lower layer, optical switching as a cheaper alternative to IP routing. “The number of [wavelength-selective] switches per node is growing quite dramatically," he says. "We are now seeing deployments with, on average, 6-8 switches per node and people are projecting as many as 20 as people start deploying colourless, directionless, contentionless-based switching."
But such demand is coupled with fierce competition among numerous players at each layer of the optical industry's supply chain.
"Some 80% of the optics used by system vendors are bought. How do you differentiate on features above and beyond what you are buying?"
Supply chain
The annual global operator market for wireless and wireline equipment is valued at US $250bn, says Schwerin, using market research and financial analyst firms' data.
The global optical networking equipment market is $15bn. The Chinese vendors Huawei and ZTE now account for 30% of the market, while Alcatel-Lucent is the only other major vendor with double-digit share. The rest of the market is split among numerous optical vendors. "If you think about that, if you have 5% or less [optical networking] market share, that really is not a sustainable business given the [companies'] overhead expenses," says Schwerin.
The global optical component market is valued at $5bn. It is likely larger, anything up to $8bn, argues Schwerin, because of the Chinese optical companies supplying Huawei and ZTE.
"You have a $5-8bn market selling products into $15bn, and then the $15bn is trying to repurpose that material and resell it to the carriers - is that really what is going on?" says Schwerin. To this vendor hierarchy is added contract manufacturers, with different players serving the component and the system vendors.
The slim profits operators are making on their services is forcing them to place significant pricing pressure on the system companies that already face fierce competition. Meanwhile, the optical component and contract manufacturers are also trying to make money in this environment.
Looking at gross margin data from Morgan Stanley, Schwerin says that the system vendors' figures range from 35% for the low end to 40% at the high end. "What the figures highlight is a lack of differentiation," he says. "And, in part, it is because they are buying all the same technology."
Schwerin says that some 80% of the optics used by system vendors are bought. "How do you differentiate on features above and beyond what you are buying?"
The optical components vendors' gross margins of a year ago were 30%. More recent data shows these figures are down, with the only segment showing a rise being optical sub-systems.
What next?
Schwerin says one way to improve the health of the industry is greater vertical integration. How this will be done - which players get consumed and how - will only become clear in the next 2-3 years but he is confident it will happen. "There are just too many layers of the ecosystem and it is just too fragmented," he says.
Operator mergers and slower spending put pressure on vendors at each layer of the supply chain, inducing revenue stalls. "These swings seems to be more and more violent," says Schwerin. "It is difficult for companies to maintain themselves in these cycles, let alone innovate."
Schwerin highlights Cisco System's acquisition of silicon photonics start-up, Lightwire, earlier this year, as an example of a system vendor embracing vertical integration while also acquiring innovation. Another example is Huawei's acquisition of optical integration specialist, CIP Technologies.
"The business model of the communication sector needs to change, and change in a relatively short order," says Schwerin, who believes it has already started. He cites the merger between the two large optical component vendors, Oclaro and Opnext, and expects a similar deal among the system vendors: "One of those 5 percenters will be absorbed."
As the market further consolidates, and as system companies drive fundamental technologies, the components' market will start to shrink. "It is then like a chain reaction; it forces itself," he says.
Schwerin's take is that rather than continue with the existing optical component and contract manufacturing model, what is more likely is that what will be supplied will be basic optical components. Differentiation will be driven by the system vendors.
Is optical components becoming a buyer's market?
"An organisation's gross margins ride on these new products"
Daryl Inniss, Ovum Components.
The global optical component market was down 2% in the second quarter of 2011 at US $1.55 billion, according to Ovum.
The good news is that the market research company is forecasting that modest growth will resume this quarter now that the build-up in component inventory that led to the market contraction has largely been worked through.
But Ovum is warning that there are signs that the continued weak market conditions and fierce competition could lead to sharp price declines even for newer, high-valued products. "An organisation's gross margins ride on these new products," says Daryl Inniss, practice leader, Ovum Components.
Oclaro's CEO on a recent earnings call said he was being asked for price concessions on 40Gbps products. Ovum also says the ROADM and tunable laser XFPs markets are becoming more crowded and competitive.
Inniss stresses that there is no evidence that companies are cutting prices to gain an edge but while he expects volumes will grow, intense pricing pressure should now be expected.
LightCounting points out that the slowdown in sales of optical component and modules in early 2011 has been limited to products that did very well in 2010 or which had long lead times, like wavelength-selective switches for ROADMs and 40Gbps modules. It says there is little, if any, excess inventory of components accumulated across the broader market.
"The telecom transceiver market remained steady in Q1 2011, but it declined further in Q2 mostly due to lower sales of 40Gig client-side modules," says Vladimir Kozlov, CEO of LightCounting. "We expect that by the end of this year, the telecom market segment will be strong again."
Best in a decade
The second quarter market dip follows a period where the optical components industry experienced its strongest yearly growth for a decade. The market reached US $6 billion for the year ending first quarter 2011 - a first since 2001.
So long as network expansion keeps up with traffic, we are looking at sustainable growth”
Vladimir Kozlov, LightCounting
The six quarters of consecutive market growth up to the second quarter was due partly to the overall health of the telecom industry. The service provider industry - wireless and wireline - grew 6% year-on-year between 2Q10 and 1Q11, to reach $1.82 trillion. In turn, the equipment market, mainly telecom vendors but including the likes of Brocade, grew 15% to $41.4 billion.
Ovum attributes the 28% growth in optical components between 2Q10 and 1Q 2011 to strong growth in the fibre-to-the-x (FTTx) market as well as new revenues entering the market from datacom players. A third factor was optical equipment vendors over-ordering long lead-time items – such as ROADMs – to secure supply.
“ROADMS did grow nicely but if you look at wavelength-selective switches, it is not such a big market," says Kozlov. The market research firm says the wavelength-selective switch market was $280 million in 2010.
LightCounting says 10 Gigabit SFP+ optical transceivers was a market highlight in 2010, with volume shipments tripling. Ethernet SFP+ sales alone reached $180 million in 2010, and will grow to $250 million this year.
“The optical component market grew 36% in 2010, and in 2011 we’re projecting it will grow 7%,”says Inniss
But competition is intense. Finisar may be the market leader but only 4% market share separates the players in second through to sixth place, says Ovum. “It’s a very competitive market and there is no breakaway here,” says Inniss.
Another challenge is the emergence of the Chinese optical component players. The large-scale deployment of FTTx being undertaken by the main three Chinese operators means that there is a huge market opportunity for local optical component and module players. The Chinese market also accounts for half the all 40 Gigabit-per-second shipments, according to Infonetics Research.
“Looking at the western suppliers, everyone is reporting slowdowns and drops in the second quarter [of 2011],” says Kozlov. “Yet from the data we are getting from the Chinese optical component players, they grew 35% in 2010 and are on track for 30% growth this year.”
Another challenge is for firms to fund sufficient R&D. Share prices took a severe hit after the companies issued warnings about second-quarter sales. “The entire optical component market is depressed because of the localised correction,” says Inniss. “It will still grow but because it is so much smaller than 2010, capital markets are bashing the companies.”
Since the stock market is an important source of investment, it may take several years for the market to recover the share price levels at the start of 2011. “It won’t stop investment in technology but there is going to be real hard eyes on each decision that is made,” says Inniss.
The main challenge facing optical component players is not so much technical issues but more the requirement to continually decrease costs. This is not new but neither is it going away, says Inniss.
Positive outlook
Yet the analysts expect market growth to continue.
Inniss points to the growing role of optics for short-distance interfaces: “The I/O (input-output) bandwidth requirements are sufficiently high, whether it is the backplane or chip-to-chip connections, that the market realisation is that optics will play a role.”
Ovum also highlights consumer market developments such as the USB 3.0 interface which will drive the market for active optical cables. “It [the consumer market] is not going to happen tomorrow - meaning 2012 - but it is something that is coming and has the potential to transform the industry,” says Inniss.
“Companies such as Finisar and Avago [Technologies] are becoming more assertive in enforcing their intellectual rights,” says Kozlov. This is as a positive development that has been missing in the past: “Protecting your intellectual property ultimately helps you become profitable,” he says.
LightCounting also highlights the need for network investment to keep track with traffic growth. "So long as network expansion keeps up with traffic, we are looking at sustainable growth,” says Kozlov. See Plotting transceiver shipments versus traffic growth.
This article is based on a piece that appeared in the ECOC 2011 exhibition guide.