If planned and engineered correctly, fiber optic systems offer high-bandwidth, flexible backbone infrastructures for demanding LAN applications. Here's a guide aimed at building and campus backbones, with recommendations that apply equally to fiber to the desk (FTTD) applications.
May 1, 2001
Optical fiber cabling in the local area network (LAN) is usually designed in one of two architectures: the single point administration or the traditional hierarchical star architecture. Together, these architectures provide a great deal of freedom for designing building backbones and closets. These architectures take advantage of fiber’s long distance and high bandwidth capabilities and provide optimal solutions to meet the customer’s needs.
To briefly review these architecture options, an example of the hierarchical star architecture is shown in Figure 1. It is a two-level star topology with cross connect facilities in both the equipment room and telecommunications closets on each floor. It provides maximum flexibility for locating LAN electronics, allowing both distributed placement in the closets or centralized placement in the main equipment room. This design is compliant with both the ISO/IEC 11801 and ANSI/TIA/EIA-568A building cabling standards.
The single point administration architecture shown in Figure 2 is a single-star topology with a cross connect located only in the equipment room. In a backbone design, all links run from the equipment room to the work area or telecommunications closet. This streamlined design optimizes the installation for centralized equipment and reduces the cost of the cabling, electronics, administration, and maintenance.
To further expand the design options, variations of the horizontal and work area subsystems, known as zone cabling, can be added to the traditional individual cable and outlet design. Both the traditional and the zone cabling options can be used with either the hierarchical star or single point administration architectures to provide a wide variety of equipment room, riser backbone, closet, horizontal and work area subsystem combinations.
Figure 3 depicts a cost-effective way of integrating both UTP and fiber media for desktop and backbone subsystems. (The UTP media is shown in dashed lines, while fiber is shown in solid lines). The design uses UTP in a hierarchical star architecture with full cross connects in the closets. This allows placement of active equipment in the closets within the 100-metre transmission distance limitation of UTP. The fiber network design takes advantage of fiber’s longer distance capability to streamline the installation by using single point administration architecture.
The UTP network provides connectivity to both the PBX and to hubs or switches in the closets for data applications running on PCs. The fiber network supports the backbone connectivity of the UTP hubs today, and stands ready to support cost effective migration to tomorrow’s data applications running on high-powered workstations or advanced network PCs.
GENERAL DESIGN CONSIDERATIONS
Each site has its own unique requirement, however the design architecture is consistent among all installations and the process should follow the procedure depicted in Figure 4:
A fiber system may be designed with specific requirements in mind or to provide a “generic” distribution system for a facility. In the first case, a significant level of detail should be available, regarding the types of devices to be connected and their associated interface requirements (this is required independent of installation method chosen). For a generic design, only architectural and environmental information may be available. In either case, the first step in developing any design is to collect as much information as possible about the site.
The end result of the data collection process should include a complete facility layout, which identifies potential node locations and marks all equipment locations, quantities and reliability requirements. In addition, a table that lists the point-to-point connectivity requirements for each type of communications interface and type of equipment must be developed (see Figure 5).
Physical design for the cabling system can be divided into two levels: high level design and detail level design (see Figures 6 and 7). The high level design addresses issues related to locating and routing the backbone; the detail design addresses issues relating to the subsystems, such as the exact location and number of cross-connects, the size of cables and the number of patch cords.
The following guidelines should be observed when locating cross-connects:
A building that supports a critical process should have at least one backbone cross-connect.
Cross-connect locations should be clean and secure. If necessary, secure equipment rooms should be constructed.
Cross-connect fields are often out on a factory floor. If so, they should be placed inside an enclosure for protection. If there is heavy traffic in the vicinity (i.e., fork lifts), elevate the cross-connect.
Avoid locating cross-connect near potential hazardous or explosive areas.
Avoid locating cross-connects near large motors or generators. Although fiber is immune to EMI, the electronics may not be.
The overall facility drawing (with potential cross-connect locations marked) should be used to lay out potential backbone routes. In many cases, this can be done by “connecting the dots”, which represent potential cross-connect locations identified during earlier walkthroughs. This process is the same whether it is an existing facility or a new building or plant.
The preliminary routing should then be reviewed to ensure that the design rules are satisfied. For example, if the backbone does not allow for redundant paths, depending on the reliability level required, potential redundant routes should be identified to enable this. The distance between the cross-connects and the main equipment room should be assessed depending upon the application to be supported. The final result should be a backbone cable route connecting all critical points either within a building or across a campus with redundant paths.
The logical design (see Figure 8) considers the applications and services required in the facility. This part of the design may not be possible in a generic design, but some level of expected applications and equipment should be assumed. The following guidelines should be considered:
Maintain protocols in native format.
Select electronics that support industry standards.
Keep separate networks/service types on separate fibers.
For topologies that support ring or secondary cable routes, separate primary and secondary paths when routing over the backbone.
For critical applications on the backbone that do not support ring or secondary cable routing, plan redundant rerouting.
Design validation should be performed to ensure that all designs meet certain parameters, for both the physical and logical requirements. Fiber types and routing should be based on application requirements, distances and redundancy requirements.
The logical validation pertains to the selection and configuration of electronics and services over the physical cable plant. Whenever there are standards for a particular service or protocol, always validate against this standard. When proprietary protocols are involved, vendor specifications are used for validation. The following is a list of issues that are usually addressed for the logical validation:
Optical parameters (wavelengths, source types etc.)
Power margin analysis
Distance between electronics
Number of stations
CABLE SIZING AND SPECIFICATION
When sizing and selecting cable, the following rules of thumb are useful. (Keep in mind that these are only guidelines):
A growth factor of between 50 per cent and 100 per cent is often applied after sizing current needs.
Factors affecting industrial backbone cable sizing include:
Campus size (number of buildings)
Level of automation (amount of computer connectivity)
Amount of point-to-point traffic vs. LAN traffic
Level of redundancy
Cable type selection is based on environment and cost.< P> Use a mixture of singlemode and multimode fiber optic cables (recommended 30:70 split minimum, respectively).
Should you install singlemode, multimode or both types of fiber? What type of multimode should you use: 50 m or 62.5 m? A few key issues guide the choice, including the intended applications support, distance, data (baud) rate, and the difficulty and expense of retrofitting at a later time.
Historically, there was not much discussion about what type of multimode fiber to install. Most fiber-based networks today deploy 62.5/125 multimode fiber in their backbones and risers and, in some cases, all the way to the workstation. However, as backbone speeds have increased, questions have surfaced about the differences between 62.5-micron and 50-micron multimode fibers. Which fiber is better?
The answer depends on the parameters of the network: the applications the network will need to support over the next few years and the length of the fiber links. The good news is that both types of multimode fiber available today offer the bandwidth to support such protocols as Ethernet, Token Ring, and FDDI over the distances specified in the application standards. The standards bodies accept both fibers, and both offer migration paths up to gigabit-level speeds. While the distance capability of traditional 50 m fiber can exceed that of 62.5 m fiber for some higher speed applications when using short wavelength lasers, this improvement does not sufficiently offset the loss of operating margin, distance, and support for legacy LANs for most customers.
Specifications for a “next-generation” multimode fiber are currently under development by the cabling-standards committees. The current draft specifications propose a new multimode fiber that has enhanced bandwidth at 850 nm, sufficient to support 10-Gbit/sec transmission up to 300 meters using a short-wavelength VCSEL technology. This will make it a logical extension of the lower-cost, higher-bandwidth systems that are already driving the LAN market.
There are also options being considered for the traditional 50 and 62.5 micron fibers to support 10Gbps using WDM (Wave Division Multiplexing) type technologies in the 1300 nm wavelength region over 300 meters, which is fast becoming the recommended maximum distance for building backbone applications. However, the question should not be about 50 or 62.5 — it should be looked at from a bandwidth, application and distance standpoint.
Multimode fiber has the capability to provide work area LAN services far into the future. In the campus backbone, however, where speeds and distances are generally ten times greater than to the desk, the clear trend is toward singlemode solutions. For the longer distance requirements of campus networks, singlemode fiber provides the LAN solution at these higher rates.
Keep in mind that fiber optic installations also require consistently low loss connections (that can be rearranged) to attach the cables. The LC connector, in particular, has been designed for high bandwidth, low loss applications such as Gigabit Ethernet, and exhibits loss performance of 0.1 dB, far in excess of competitive products and connector standards.
If planned and engineered correctly, fiber optic systems offer high-bandwidth, flexible backbone infrastructures for the LAN applications of today and tomorrow. With the growing amount of data traffic now flowing across networks, the backbone becomes and increasingly important part of the cabling infrastructure. Be sure to avoid the potential bottlenecks.CS
James Donovan is Strategy & Development Director for Connectivity Solutions in EMEA at Avaya in Bray, Irelend. He originally joined Lucent Technologies in 1993 as Technical Manager UK & Ireland region. Before joining Lucent, James worked at GEC, ITT and Alcatel, where he was responsible for developing and supporting cabling and data networking equipment.