Anaheim, Calif.—Anyone who uses fiber optics for communications uses pluggable optical modules to connect data-carrying optical cables. Large datacenters may have more than one million such modules. As a result, users of these optical transceiver modules that plug into switches and routers always want higher speed, lower power, and of course, lower cost. Nobody wants to pay more, even as data rates multiply.
The profitability crunch was the topic of a panel discussion, "The Promising Market of 100G Pluggable Devices" held on March 22 at OFC 2016. Representatives of pluggable module users, module makers, and their suppliers spoke on why, they claim, module makers are making many modules, but not making much money. "The optical community is made up of the smartest people doing the stupidest things to not make money," said Vipul Bhatt, director of systems engineering at Inphi.
Vipul Bhatt from Inphi addresses the audience at OFC 2016.
As datacenters transition to 100 Gbps optical links and look toward 400G, they will be replacing their line cards to achieve higher port densities and higher throughput, and that can result in new cables and new pluggable modules. The panelists cited several reasons why today's array of pluggable modules aren't all that profitable.
QSFP is just one of several pluggable optical modules that can handle 100G.
"There are too many MSA (multi-source agreements) and thus, too many variations of pluggable modules," said Kenneth Jackson, product marketing director at Sumitomo Electric. User demand for smaller sizes has led to a wider range of pluggable modules that, according to Jackson, has raised costs. The industry has yet to standardize of a form factor for pluggable optical transceiver modules, though the QSFP seems to be in the lead. Even so, others such as micro QSFP and QSFP28 have been designed to handle 100 Gbps.
This fiber-optic cable uses a microQSFP pluggable module, on display in the TE Connectivity booth at OFC 2016.
QSFP28 is another pluggable module contender for 4x25 Gbps optical links.
"QSFP28 will trigger another merry-go-round of MSAs," warned Matt Traverso of Cisco. This, according to the panelists, will lead to more form factors and higher, not lower, costs. Jackson noted that "packaging engineers have done too good a job," which led to the creation of so many form factors. Traverso also blamed the writers of the MSA's for increasing costs saying "MSA's don't initially have all the details. It's not until system integration do the missing details arise."
Another reason cited for low costs is there are too many companies making pluggable optical modules. That sounds like a good deal for those who use them. Yves Le Maitre of Octaro said "there is no room for "me too' products. You'd better be good at what you do to make money in this business."
Le Maitre also argued that the speed at which the technology is moving results in higher development costs. "The technology has moved faster than we predicted a few years ago, he said."
Confused? Welcome to the club.
The datacenters are clearly driving the pluggable optical module market. Volumes are huge, which means economies of scale should result in lower costs for manufacturers. Because it's a buyer's market, the buyers are demanding whatever it takes to lower their costs. Because of that market pressure, the suppliers claim they can't make money and one of the reasons they cite is too many suppliers? How can there be too many suppliers if there is enough of a market to support the existing number of suppliers? It seems to me that the market is working as it should. If profit margins were too low, suppliers would exit. So, pluggable optical-module suppliers, if you're margins are too low, why are you still in the market?
The module makers also blame their customers for wanting what works best for them and not necessarily using what's an industry-standard product. Can you blame your customers for wanted what works best for them at a low price?
Bhatt said that "For the next five years, pluggable optics will remain alive. He might be right, but there's already a movement to replace them, which was the topic of another panel at OFC 2016. Unfortunately I couldn't attend that session.
—Martin Rowe, Senior Technical Editor