In a space where customers don’t always react favorably to change-as fiber proponents who have attempted to wean customers off their copper dependency can testify-the premises cabling market enters 2005 with a new favorite fiber, a new application to exploit, and two new standards that should either see or approach ratification within the next 12 months. These changes enjoy a fair amount of synergy; the new lead fiber, laser-optimized 50-µm multimode (MM), should find a prominent role in the two new standards, and one of those standards is aimed at data centers, the burgeoning application. Industry sources pick these factors as positive omens for the extension of fiber into more campuses and enterprises. Yet, they also agree that unless they do a better job of educating their potential customers, the more things change, the more they’ll stay the same.
While laser-optimized MM fiber has been part of many cabling portfolios for at least two years, it has only been over the past year that this technology has gained traction in the marketplace. As detailed in an earlier article (see “Laser-optimized multimode finally gains some traction,” September 2004, front page), the latest flavor of 50-µm MM was codified in the IEEE’s 10-Gigabit Ethernet standard as the only current MM option capable of accommodating 10 Gbits/sec over at least 300 m via single-wavelength transceivers. (The 10GBase-LX4 PMD offers an option for legacy 62.5-µm fiber but requires a four-wavelength transceiver.) While it has benefited from an overall trend away from 62.5-µm cabling, laser-optimized plant still represents a small percentage of overall sales.
However, that percentage is expected to grow, particularly as 50-µm MM in general becomes more popular. According to Sandra Young, vice president of marketing, network products, at Commscope (Claremont, NC), industry research indicates that 50-µm MM has grown from 3% of shipments three years ago to more than 30% today. Adds Commscope’s Stephen Bailey, market manager, network optical products, the decision to move from 62.5- to 50-µm fiber generally depends on how quickly the network designer expects to require support of data rates approaching 10 Gbits/sec.
Commscope is a firm believer in the laser-optimized technology; it’s now the company’s standard 50-µm MM offering. However, other cabling companies offer both laser-optimized and “regular” 50-µm products. One reason may be price; laser-optimized fiber can be anywhere from 20% to 75% more expensive than vanilla 50-µm, depending on whether premium laser-optimized varieties that can support distances of 500 m or more is being used as a point of comparison.
However, both types of 50-µm MM are more expensive than singlemode (SM)-until you add in the cost of electronics. So for users who envision high speeds and runs in ranges of about 300-500 m, the total infrastructure cost of laser-optimized MM will still compare favorably to SM.Yet even with 50-µm MM gaining popularity, 62.5-µm cabling won’t be going away anytime soon, Young and Bailey suggest. First of all, inertia tends to lead some network managers to extend their installed base of 62.5-µm MM when adding new users, particularly if 10-Gbit/sec data rates remain well over their planning horizons. Technical limitations also can come into play; while transitioning to 50-µm is easy and economical when the new plant is electronically connected to the old via a patch panel, spliced additions are not.
Another incentive to stick with 62.5-µm MM may come with the advent of the 10GBase-LRM transceiver PMD, which promises to offer a way to support 10-Gbit/sec rates over the legacy MM via serial transceivers. Young and Bailey as well as other cabling industry sources say customers have little or no familiarity with the LRM standards work and therefore aren’t basing cabling choices on having LRM devices available to meet their future requirements. But with LRM-based devices expected to be announced this year, it’s safe to bet that transceiver manufacturers will work diligently to change that.
One standards-making activity of which network managers are likely more aware-and which may further boost the demand for laser-optimized 50-µm MM-involves an update to the venerable TIA-568 commercial cabling standards. The upcoming version, TIA-568C, is not expected to be ratified until June 2006. However, draft versions should be available this year that may put laser-optimized cabling front and center when it comes to recommended cabling types. It also provides an opportunity to address technical complexities the industry has fostered in the name of choice-although observers doubt this opportunity will be fully ceased.
Unlike previous versions of the 568 standard, which have treated optical technology-some would say slightingly compared to copper-in bits and pieces throughout the document, 568C will collect all the optical recommendations into a single section, 568C.3. According to Paul Neveux, director of premises cable product management at Superior Essex (Atlanta), concentrating the optical information into a single section will make it easier for network designers to transition from copper to fiber. “It contains all of the optical-fiber information in one location,” he says of the expected standard. “So it’s going to be a lot easier for folks to find the information that they need to be able to apply it to data centers or the enterprise space.”
Industry observers agree that the standard will include laser-optimized 50-µm MM in its list of “approved” transmission media. But while Neveux expects the standard to be essentially neutral in the emphasis placed on 62.5-µm, standard 50-µm, and laser-optimized MM (not to mention SM), others aren’t so sure. For example, the new TIA-942 standard for data centers says that while 62.5- and 50-µm cable that meets TIA-568B specifications are acceptable for data centers, laser-optimized fiber is recommended. “Based on this, I would not be at all surprised if they recommend laser-optimized in 568C for the fiber part of the cable plant,” offers John Struhar, director, fiber structured cabling system solutions, Ortronics (Smyrna, GA), and president of the TIA Fiber Optics LAN Section.
“I think it’s going to move to [laser-optimized] 50-µm,” agrees Commscope’s Bailey. “There’s no reason not to go there. Your legacy business in the United States is 62.5; if you’re going to make a change, it ought to be worth something. That’s why we’ve made the decision to only offer the laser-optimized product.”
But fiber type isn’t the only technology debate that will likely be on the table during the 568C deliberations. One topic already raised is the possibility of settling on a single small-form-factor connector, rather than the current approach of saying that any connector that meets a certain set of specifications and earns a FOCIS document can be called “standards-compliant.” The Fiber Optics Association has already attempted to insert its own connector design as a “neutral” approach toward this goal. While that effort has failed, it’s not because the concept behind it is flawed, sources say.
“I think the market would like to see a single standard connector,” Neveux says. “But even though they’re raising it while they’re editing this particular document, I don’t think they’ll settle on an LC connector, for example, and say, ‘Okay, this is what we recommend as our standard small-form-factor connector.’ ”
Struhar agrees: “I don’t know if it will be as soon as 568C comes out, but I do feel that ultimately the industry will gravitate toward one small-form-factor connector, with or without pushing from the standards people. Performance is critical, and if other connectors don’t meet the requirements, then people are going to move toward connectors that do. So I think there’s a good chance it will happen, but I think that it’s still too early for the standards committee to force it.”
As Struhar implies, the impetus for change would appear to lie outside of the standards process. “Here’s where I think the influence is going to happen: It’s going to be with the box manufacturers,” predicts Neveux. “It’s going to be the guys like Cisco and Nortel and Lucent and whoever makes the Ethernet switches and the boxes. I think it’s going to be difficult in a standards arena, since all of the [connector] manufacturers are represented.”
While ratification of 568C is at least a year away, the TIA-942 data-center cabling standard should be ready for use early this year. The timing will be excellent, since data centers, traditionally copper enclaves, are now facing the prospect of the high-speed data rates that have spurred optical equipment growth in the enterprise.
As mentioned, TIA-942 references TIA-568 when it comes to cabling. It also echoes the commercial cabling standards in recommending a hierarchical network structure and centralized cabling architecture. Thus, many expect data-center designers to begin thinking more like their enterprise counterparts. “When the draft standard 942 is ratified, you’re going to be putting these designs on somewhat parallel tracks,” says Struhar of data-center and enterprise applications. “The data-center draft standard allows a hierarchical architecture just as 568 does. When that’s ratified, I think we’re going to start seeing people thinking more in terms structured cabling in data centers than they have been in the past…With the standard, we should get some uniformity-and with the uniformity, I think we should get people starting to think about fiber types similarly.”
Uniformity is not something common when it comes to data centers now. “It varies all over the board,” Struhar says of the current data-center landscape. “Some large customers are still cabling their data centers with 62.5-µm fiber. Others are using 50, and still others are putting in laser-optimized. So it really depends. And there doesn’t seem to be a correlation with the size of the data center as to what fiber type they’re using. Some very well-known companies are still putting 62.5 or conventional 50 in, whereas some other companies are going with laser-optimized.”
That said, there are traits characteristic of data centers. One is the reliance on copper. “I think mostly because of the familiarity with the medium,” Neveux offers. “The people that they have in the field, the technicians, they need to get a data center up really quickly, so the way to do that is that they use standard cabling techniques. So they basically structure their data center around a copper configuration.”
Another common trait is short cable runs, with typical maximum lengths of 100 to 150 m, Struhar says. The typical data-center configuration has much to do with this, Neveux adds. “Everything is concentrated. It’s almost like a central office, in a way-not quite-in terms of you have all of your electronic equipment in bays and you’re basically bringing all of your cabling into a centralized area,” he explains.
Fiber should have some natural advantages in such a crowded data-intense environment, Struhar and Neveux say. First, fiber cables are smaller and generate less heat than Category 6 or coaxial cable, two important considerations in an application where space is cramped and cooling issues are both significant and costly. Second, fiber’s natural data-rate advantage can be extended because the shorter cabling runs enable engineers to trade off loss characteristics to accommodate more complex architectures or optical components.
For example, two of the larger loss components in the IEEE loss models are intersymbol interference (ISI) and insertion loss. The former is distance-related; since the distances in a data center are shorter, the loss the model budgets for ISI can be reallocated to insertion loss. That can mean the accommodation of additional single-fiber connectors as well as either ribbon connectors or cable assemblies.
The latter two technologies should prove popular in the data center. While ribbon connectors such as the MPT/MPO almost have to be factory-terminated (Corning has released a field-terminated connector), both Neveux and Struhar see pre-terminated cable assemblies using single-fiber connectors as natural products for data-center applications.
“I think that’s going to be a prime market for pre-connectorized cable,” says Neveux. “Because it’s going to be a setup much like a central office as far as all the equipment being in bays and well organized. And they’re going to have crossconnects that are going to require jumpers and they’ll have standard lengths and standard procedures that they use to install that.”
Of course, the problem with entering a new market is introducing yourself to people who don’t know you-or are leery of your optical technology. Since customer education has been something fiber proponents have faced in the enterprise, that won’t be anything new. Unfortunately, since that process also remains unfinished in the enterprise, vendors are fully aware that they have to improve their efforts if they want to succeed in data centers.
“I think there’s a reluctance on the customer’s part to just jump into the fiber arena, only because they’re not familiar with it, they’re not comfortable with it,” Neveux concedes.
“There’s still this fear of fiber,” Commscope’s Young laments. “It’s changing….The contractors and installers from age 25 to their 30s seem to be less fearful of it and quicker to make these decisions to do whatever they need to do.”
However, unless network designers old and young are exposed to the benefits of optical infrastructure in a language they can understand, new standards and new technology won’t open new markets, no matter how appealing they may appear.