An evolution of optical cabling for the data center has taken place in the last few years, with eight-fiber-based solutions becoming popular for optimizing the structured cabling to the transceiver roadmap. As this shift continues, it’s helpful to review the industry drivers and infrastructure challenges that led to the development and popularity of eight-fiber-based (or base-8) cabling. By looking back as well as into the future, we can evaluate base-8’s market adoption and its value.
To understand how the base-8 solution was born, we must look into the history of data center cabling. Veterans of the cabling industry may recall the painstaking days of field-terminating and installing fibers, one at a time. As the number and size of data centers grew, designers and installers had to manage hundreds or even thousands of single- and dual-fiber connectors. But with space and time being premiums in the data center, this approach created challenges in both the density and speed of deployment.
The MTP (MPO-style) connector, launched in 1996, revolutionizes the design and deployment of cabling in the data center. Having 12 fibers in a single ferrule, it made dramatic progress in meeting the high-level challenges of speed, density and ease of installation.
The first common question is when and why base-12 solutions were the first to see wide deployment. Base-12 connectivity was introduced in the mid-1990s, driven by a desire to develop a modular, high-density, structured cabling system that allowed quick deployment while also maximizing port densities in the rack. The TIA/EIA-568A fiber color-coding standards are based on groups of 12 fibers, and ribbonized fiber was typically in groups of 12. Basing high-density connectivity on an increment of the number 12 made sense, so the 12-fiber MTP connector—and base-12 connectivity—was born.
Historically, MTP-terminated trunk cables have served as the backbone for data center structured cabling, from the central patching area out to zone or equipment-distribution areas. The primary data rates at the time didn’t exceed 10GbE, so the optical ports on the servers, switches and storage devices were duplex, or dual-fiber based. With high-density base-12 backbone connectivity through MTP connectors, a base-12-to-base-2 (MTP to LC duplex) breakout module or harness provided the dual-fiber interface for the dual-fiber port at the equipment. Since the number 12 is wholly divisible by the number 2, this MTP cabling and breakout solution could easily provide the dual-fiber interface into the network equipment with full fiber utilization of the base-12 backbone trunk cables. In fact, using the MTP connector for the structured cabling enables the deployment of six duplex ports with a single connector mating into a module.
Fast-forwarding to 2009, the data center infrastructure challenges were growing, and density and speed of deployment became even more critical. Before this time, most data center solutions were based on hardware designs that commonly served in the traditional LAN market. The need to solve the problems specific to the data center emerged—not just generic cabling problems. As the data center market continued to grow, it was clearly ready for a purpose-built pre-terminated fiber-optic solution. When such a solution arrived, it addressed the challenge of consuming costly data center space for cabling while also optimizing the infrastructure components for ease of use and speed of deployment. The value it provides to the industry continues to be proven almost 10 years after its market release. Density, network uptime, speed, simplicity and a clear transition path to meet future requirements were its foundations.
From 2009 to 2013, optical technology and protocol roadmaps associated with data centers continued to evolve and crystalize. Discussions with major transceiver, switch, server and storage manufacturers clarified that many transceiver-technology choices would emerge, but all would be based on either dual- or eight-fiber connectivity. In other words, for Ethernet transmission ranging from 40G to 400G, all roads lead to these solutions. The table below shows this situation.
Blue font represents IEEE standard protocols as of 2017.
Although dual- and eight-fiber solutions are the future, a few short-lived alternatives will offer a different base fiber count. As the table above indicates, the road to 400G includes proposal of the first- and second-generation OM3/OM4 parallel transmission as base-32 and base-16 solutions. Discussions with prominent transceiver, switch, server and storage vendors, however, indicate these solutions won’t see wide deployment owing to manufacturing cost and connector complexity.
Imagine a backbone infrastructure using an MTP connector with a 16-, 24-, or 32-fiber increment. Introducing this connector into the cabling infrastructure means you’re cabling for the least common denominator: the transceiver technology that will serve least frequently and have the shortest life. With an infrastructure based on something higher than 8 or 12 fibers, a form of down-conversion will be necessary to interface with more-common parallel transceivers, such as SR4 or PSM4. Conversion modules, while enabling interface to optical ports with different fiber counts, increase cost and link insertion loss. The third-generation solution for 400G parallel transmission over OM3/OM4 fiber, an eight-fiber solution, is expected to garner widespread market acceptance. Since the number 8 is wholly divisible by the number 2, base-8 backbone connectivity can easily work with 2-fiber transceivers, just like base-12. But base-8 connectivity provides the most flexibility for what are expected to be the most common 40G, 100G and 400G transceiver types. Simply stated, base-8 connectivity provides the most future-ready solution for 10G to 400G transmission.
Technology changes happen quickly in the transceiver world, but anyone who has installed 40G circuits will know that a common transceiver type is QSFP, which often employs eight fibers. Base-12 can connect to 8-fiber QSFP ports; indeed, many people operating 40G circuits today have installed base-12 connectivity in their backbone. Clearly, however, installing a 12-fiber connector into an 8-fiber transceiver means that four fibers are unused.
To resolve this issue, vendors released solutions that enabled 100-percent utilization of the backbone fiber when employing base-12 cabling with parallel base-8 transceivers. This feat involved base-12-to-base-8 conversion modules or harnesses. But solving one problem only introduced another: conversion modules add insertion loss because the link has more MTP connector pairs. This situation hinders link performance and increases the cost associated with a conversion module, the industry needed a better way.
The solution to managing a world based on 2- and 8-fiber technologies is base-8 connectivity. It required new pre-terminated cabling designed for the data center—this one based on eight-fiber technology. This solution provides all the value of the original pre-terminated solutions, with the addition of superior network scalability, improved link performance and 100-percent fiber utilization. Base-8 solutions bring additional value across various networking applications as well as easy management of cabling infrastructure.
New networking applications have emerged, such as breaking out higher-speed ports (40G) to lower-speed ports (10G) to decrease cost and increase density. Port-breakout deployments have become a popular networking tool and are driving the large industry demand for parallel optical transceivers. Today, port breakout commonly serves when operating a 40/100G parallel optics transceiver as four 10/25G links. Breaking out parallel ports is beneficial for multiple applications, such as building larger-scale spine-and-leaf networks and enabling high-density 10/25G networks.
How does this application relate to the new base-8 cabling? Just like deployments of native 40G or 100G networks, the parallel transceivers operate over eight fibers when serving in port-breakout mode. The base-8 cabling provides an exact match to the transceiver requirements, eliminating unused fibers.
In addition to addressing fiber utilization, link loss, costs associated with deploying high-density and high-bandwidth networks, and other issues, one final challenge for base-8 is the complexity of managing a cabling infrastructure where the MTP connector plugs directly into the transceiver. Historically, cabling solutions have used pinless MTP-to-MTP trunks for duplex tasks such as 10G. These trunks plugged into MTP-to-LC breakout modules, and LC duplex jumpers patched from the module to the data center electronics. As MTP pairs require a pinned-to-pinless connection, the MTP-to-LC modules employed a pinned MTP connector inside the module. When moving to a parallel optical system, the trunks are then installed into MTP adapter panels, and MTP jumpers connect from the trunks to the electronics. This is where cabling challenges can arise.
Since parallel optical transceivers are always pinned, a pinned-to-pinless jumper is required to connect the pinless MTP-to-MTP trunk to the transceiver. As depicted below, depending on the link, two additional jumper types are necessary: a pinned-to-pinned MTP jumper for connecting two trunks in a main distribution area and a pinless-to-pinless MTP jumper for directly connecting two transceivers. These configuration requirements triple the quantity of jumper varieties compared with traditional duplex systems that use a single LC duplex jumper. This jumper complexity also introduced risk to the data center, potentially causing damage if the wrong jumper was at the wrong location.
To solve these issues, the new base-8 solutions feature pinned MTP-to-MTP trunks. This change allows the use of a single pinless-to-pinless jumper throughout the fiber infrastructure regardless of cabling design, eliminating the need for three-jumper configurations and eliminating risk.
As the strong value and linkages to the optical technology and protocol roadmaps evince, base-8 solutions have gained wide acceptance across the globe and across all verticals. The most common response from customers is why hasn’t an 8-fiber-based solution been available all along? They are excited to see solutions that are rooted in the industry’s technology roadmaps and that give them the utmost in infrastructure flexibility. Data center operators and owners know that with a base-8 solution, they’re installing cabling that minimizes their risk and is optimized for today and tomorrow.
About the Authors
Jennifer Cline is program manager for Corning. She has previously held positions in engineering services, marketing, market development, field sales and product-line management. David Hessong is manager of global data center market development for Corning. During his career with the company, he has held positions in engineering services, product-line management and market development.