Is 2020 the year of 400G? It’s the kind of question telecom industry observers love to ask, though it is a far too general one. It speaks to our fascination with big numbers as milestones to be reached and put in the rearview mirror as we push toward the next milestone.
In reality, 400G adoption and usage is guided by more practical considerations. Deployment decisions depend on the type of company adopting the technology, the current state of that company’s network and its existing equipment, and most significantly, how that company wants to use the technology and at what distance—all of which impacts cost.
The 400G Evolution
The industry’s 400 Gigabit capabilities did not just magically appear at the beginning of 2020. The first coherent 400G wavelengths started shipping in 2017, to scant demand, and through last year 400G wavelengths still accounted for only about 3-4% of the total WDM market revenue of about $14 billion, according to Jimmy Yu, vice president and analyst for the optical transport market at Dell’Oro Group. Yu this Wednesday will moderate a panel, “Next Gen Optical: 400G, 800G and Beyond,” as part of the FierceTelecom Blitz Week free virtual event.
However, as companies have started to eye investment in 400G line cards in recent years, they did not necessarily have the goal of running 400 Gbps channels. As noted by Yu and Jonathan Homa, senior director, ECI portfolio marketing at Ribbon, early 400G adoption has focused more around multiplexing multiple lower speed signals, either 100G or 200G.
“You can do a 400 Gigabit signal as one carrier, or you could do it as two carriers at 200 Gigabits,” Yu said. “Up to now, a lot of times when people would buy a 400 Gigabit capable system or line cards, they would run it at 200G.” Yu said 100G and 200G wavelengths together accounted for about 85% of the WDM market revenue last year.
Homa added, “Today, or up until very recently, the highest speed client would be 100 Gigabit Ethernet clients. What customers often have been using 400G for is to multiplex multiple 100G clients onto a 400G line.”
The two primary customer segments for 400G thus far have been data center operators looking to interconnect two or more data centers, or service providers that want to build metro rings consisting of 400G lines on which they then would lease capacity to data center operators or enterprises through a data center interconnection (DCI) as-a-service offering.
RELATED: Report: Ciena, Cisco and Infinera leaders of the Q1 DCI pack
Yu said hyperscale data center and cloud operators and MSOs have been the most aggressive in embracing 400G. Ribbon, for its part, largely caters to service providers, particularly in Europe.
Homa said that with the arrival of 400 GBE clients on the market, 400 Gigabit single-carrier channel opportunities will expand. “Next month, we’ll be going live with one of our European customers transporting 400 Gigabit Ethernet over a 400 Gigabit line,” he said. “That eventually will become the main application for 400G.”
For now, Yu said the current metro DCI sweet spot for 400G on early generation line card is 300 to 400 kilometers, but that new-generation cards will advance capabilities. Homa described 800 km as the maximum distance for what he termed “performance-optimized” deployments of 400G wavelengths. Meanwhile, dual-carrier 200G deployments could allow a 400G signal to be used for “several thousand kilometers,” he added.
Compounding the considerations is the fact that many service provider networks have spans that still use relatively narrow 50 Ghz fixed-grid spacing. About 40% of the spans in North America alone fit that description, said Sam Lisle, business development director for optical transport at Ribbon.
RELATED: Windstream Wholesale and Infinera fire up 400GbE trial
“Some service providers have constraints they need to deal with, brownfield networks that can’t tolerate a 400 Gigabit optical wavelength to carry a 400G client,” he said.
That’s another reason why companies may opt to use something like what Ribbon called the “performance-optimized,” dual-carrier 200G option. Otherwise, they may need to invest in flexible grid ROADMs to take 400G farther. Investments in flexible grid equipment and other gear, like amplifiers, that extend reach will add more cost to 400G upgrades beyond the necessary line card investments.
400G ZR’s Pending Arrival
Expense is not a trivial issue for companies that would seek to deploy 400G. Aaron Werley, chief technology officer at Zayo, said his company is deploying 400G-capable platforms. However, widespread 400G usage may require new transponders that will effectively increase range without requiring recurring, costly signal regeneration.
Brad Booth, principal network hardware manager for Microsoft Azure, said power and cost challenges are limiting 400G’s immediate term viability for use inside data centers. Booth said in an email, “While 400G offers four times the bandwidth of 100G, current implementations are at least three times the power and cost. Where 400G does make sense is in locations where that added bandwidth is worth the slight reduction in power and cost.”
Many companies eyeing 400G deployment may be waiting for widespread availability of pluggable modules based on the 400G ZR standard.
400G ZR and the longer-distance 400G ZR Plus will bring interoperability, and with that, potentially lower cost to 400G deployments as companies deploying 400G have greater options to mix and match different vendors. “Microsoft has been very public about our need for 400ZR for our regional data center interconnect links as a precursor to our deployment of 400G inside the data center,” Booth said.
But. 400G ZR and ZR Plus are likely to remain largely in a trial and demonstration phase for this year, said Ribbon’s Homa, with commercial availability ramping up next year.
That may be just fine for some network operators.
Werley, who said he’s interested in seeing what 400G ZR can do, also said Zayo’s customers are not exactly breathing down its neck demanding 400G yet. “They want to know when it will be available to them, but it’s not a requirement yet,” he said. “They’re kicking the tires.”
If the 400G market is developing at a gradual pace, it may seem early to talk about the next milestone on the optical transport technology roadmap. However, a small number of companies already have announced 800G product plans, and 1.2T is also on the horizon.
Network operators may not be talking about 800G or 1.2T yet -- Werley said 800G is not “getting a ton of focus yet” at Zayo -- but Ribbon’s Homa said the technology exists today to drive beyond 400G. “By using two 600G carriers, we can already do 1.2T today. We can carry three 400G Ethernet clients today for a 100 kilometer distance, two for 800 kilometers and one for several thousand kilometers.”
He added that making higher speeds viable in the future means “riding the DSP technology curve” and making other advancements to lower the cost of achieving these speeds at distances that network operators require.