Upcoming Events

Where the Cloud Touches Down: Simplifying Data Center Infrastructure Management

Thursday, July 25, 2013
10:00 AM PT/1:00 PM ET

In most data centers, DCIM rests on a shaky foundation of manual record keeping and scattered documentation. OpManager replaces data center documentation with a single repository for data, QRCodes for asset tracking, accurate 3D mapping of asset locations, and a configuration management database (CMDB). In this webcast, sponsored by ManageEngine, you will see how a real-world datacenter mapping stored in racktables gets imported into OpManager, which then provides a 3D visualization of where assets actually are. You'll also see how the QR Code generator helps you make the link between real assets and the monitoring world, and how the layered CMDB provides a single point of view for all your configuration data.

Register Now!

A Network Computing Webinar:
SDN First Steps

Thursday, August 8, 2013
11:00 AM PT / 2:00 PM ET

This webinar will help attendees understand the overall concept of SDN and its benefits, describe the different conceptual approaches to SDN, and examine the various technologies, both proprietary and open source, that are emerging. It will also help users decide whether SDN makes sense in their environment, and outline the first steps IT can take for testing SDN technologies.

Register Now!

More Events »

Subscribe to Newsletter

  • Keep up with all of the latest news and analysis on the fast-moving IT industry with Network Computing newsletters.
Sign Up


FIBER OPTIC DESIGN

Fiber Optic Design Considerations

Before fiber optic networks can be constructed, they must be properly designed and, once constructed, they must be managed. Efficiencies in these processes translate into lower cost layout and construction, more productive system migration and field operations, lower optical loss budget, and greater business profitability by bringing fiber to the desk.

The fiber optic network layout design plays an important role in error-free system reliability. Choice of the proper type of network layout depends on the type of process controlled, the possible need for expansion, and the degree of failure immunity desired–all of which must be balanced with cost considerations.

Basic Layout Network Designs

As is the case with electrical control system networks, four basic optical networks prevail: bus, star, ring, and collapsed backbone. For each type, the purpose of the network is to provide communication between the devices, or nodes, in the system. "Node" is a general term that refers to a programmable logic controller (PLC), remote input/output (I/O) drop, distributed control system (DCS) controller, or any communication device.

Network Layout Types

Each of the four network types has advantages and disadvantages, depending on the application. Historically, bus layouts have been preferred by PLC suppliers and star layouts by DCS suppliers–both on equipment OEM-configured for fiber optic signal transmission. Recently, both PLC and DCS suppliers have begun to offer more ring and collapsed backbone layouts than in the past.

Bus Network

In a bus network, all the nodes are attached in a line. This layout lends itself especially well to automobile assembly lines, lumber and paper mills, and other operations that begin with raw materials at one end, and end at the other end of a production line with a finished unit. The control devices are laid out in a linear array alongside the process machinery. Because the process machinery requires adequate clearances and right-of-way, it is usually easy to set up the cableways along the same right-of-way.

In an electrical bus signal transmission system, devices are connected to the nodes in parallel by direct attachment or attenuating taps. Attenuating taps allow higher speeds and greater bandwidth, but do not provide for easy network expansion.

In a fiber optic system, fibers cannot simply be paralleled as can copper conductors. Present fiber optic technology does not provide for effecting the equivalent of an electrical tap. In a fiber optic system, taps are effected through modems. All intermediate modems in the string (with the exception of the two at the extreme ends of the bus) are repeaters that interface with their respective nodes and send the optical signal on to all other modems on the bus.

Fiber optic systems are normally used in star wiring patterns or to connect wiring centers. Modems convert digital signals to analog signals and vice versa. They are normally used on analog circuits.

Star Network

The star network consists of a star (central) node device with arms extending out to other nodes. The star is used predominantly in facilities where different processes are physically separated, but must be centrally controlled, such as in petroleum refineries, chemical and pharmaceutical plants, and power-generating stations.

Outlying nodes handle individual complex tasks, or have many alternative paths, and must therefore function somewhat independently. For example, one node might be connected to thermocouples with slow temperature changes in a 4- to 20-ma control loop; a second node might be a high-speed remote computer on a 10 Mbps RS-485 link; and a third node might be a controller operating a motorized process control valve [1].

Because the controlled functions might have no well-defined path of their own, or the path might not readily lead back to the central devices, locating the cable ways and securing right-of-way can be more complex than with linear layouts. Also, in a star network, the central device is always a repeater, capable of transferring communications from one separate node to another. Sometimes the central node has overall control over the separated nodes. Each separated node is connected to the central device by a point-to-point link. Because each node receives and sends messages solely with the central device, only these two devices must understand the message. Thus, the different nodes can communicate at different speeds and use different protocols or languages. If the star node has enough power and intelligence, it handles many different speeds and protocols. This feature makes it easier to use devices utilizing various technologies from different manufacturers.

Furthermore, the central node is usually a hub or a multiplexer that utilizes repeaters to forward data. Some repeaters can interconnect cable segments using different physical media such as coaxial cables and fiber optic cables.

In a star network, however, each separated node requires two modems–one attaching to the star and other to the node for a point-to-point link. As a result, a star requires twice as many modems as a bus layout, and thus costs more.

Note:

Fiber optics termination does not utilize modems. Cables are terminated in fiber optics patch panels, medium attachment units (MAU), fiber optic hubs, etc.

Ring Network

Devices in a ring layout are connected in a circular fashion. Each node is a repeater, and all nodes operate at the same speed under the same protocol. In theory, a simple ring handles complex processes but, in practice, relatively few processes lend themselves to a ring layout. A ring network can, however, be advantageous in high-reliability applications, because it can be installed in a modified self-healing configuration.

Note:

With regard to FDDI, fiber optics is normally used to connect two wiring systems or in a star wiring plan.

Collapsed Backbone Network

Today, networks are composed of broadband, baseband, and fiber cables. Broadband is the same type of cabling used by Cable TV companies, the cable carrying multiple channels. Baseband is the typical ethernet cable used to connect machines, usually known as thicknet or thinnet. Fiber, of course, is cable composed of glass fibers and the signals are transmitted as light. This composition of multiple media makes the network environment hard to manage and the transition to future technologies difficult.

Within network environments today, there exists what has come to be known as the collapsed backbone. The idea behind collapsing the backbone is to bring all the inter-building connections into one location. This makes management easier and a chance of a cable plant problem affecting multiple buildings less likely. All the interconnect equipment usually resides in building with a controlled environment and backup power.

The daily management of the network will be greatly enhanced when the backbone is collapsed. Things that should be simple, like tracking down a bogus or bad host, are difficult in the environment today. Once the backbone is collapsed, network problems will be isolated to just one building or area. The TCP/IP topology of network environments today is only about 20% routed and the rest bridged. In the future, collapsed environment of the TCP/IP topology will be 100% routed. This will greatly increase network performance for users.

Layout Expansion Design Considerations

To expand a bus network, one adds to either end. Expansion is easy if growth is linear, but difficult if links must be added between the ends or on a branch.

The star is expanded by adding more arms with separated nodes, and their connecting cables. As long as cableways are available and the central device has enough capacity, expansion is straightforward. Moreover, it is possible for units to be added while the network is up and running.

In a ring layout, expansion is difficult. Because any addition requires disrupting the ring, it cannot be done quickly or while the network is running.

Achieving High Reliability

Studies show that in fiber optic telephone systems, 80% of interrupted service is due to cable damage. Much the same might be true in industrial environments, where cable is exposed to potential damage from sources such as forklift trucks, dropped tools and equipment, and cutting and welding torches.

In a bus or ring, the entire network usually goes down if a cable is damaged, because the network devices usually have neither the power nor the intelligence to operate as isolated entities. And in the rare event that a modem or repeater fails, communication is also disrupted throughout the network.

The self-healing ring network can be installed in a modified, ultrahigh-reliability configuration. The ring is made to send signals clockwise and counter-clockwise by duplicating the cable and installing two optical transmitters/ receivers at each node. Importantly, the two cables can be strung alongside each other in the same cableway, because operations are not disrupted if either or both cables are damaged. If one or both cables between any two devices is damaged, communication is disrupted at that point. However, the nodes adjacent to the break continue to receive communications from either the clockwise or the counter-clockwise signal stream. Likewise, if one node fails, communication continues among the other nodes.

A modular fiber optic design can bring down the cost of a self-healing ring layout. That is, rather than duplicating the entire modem, one need add only a transmitter/receiver module and a self-healing ring module to each modem. If regular modems are already in place in a bus layout, the network can become self-healing by connecting the two ends and inserting the additional modules. The ability to insert additional modules, rather than replace modems, also reduces installation time.

In practice, different parts of a network require varying levels of reliability. The most critical processes can be arranged in a self-healing ring, and less critical processes in a bus, star, collapsed backbone, or other hybrid configuration.

Note:

Fiber optics are normally installed in a star configuration and/or used to connect two wiring systems–other than the FDDI hub topology.

Many hybrid variations on the basic network types are possible, incorporating the star in one form or another. If a cable is damaged in a star, communication stops only with the node served by the damaged cable; the other nodes continue to operate. It must be borne in mind, however, that if the star (center) node itself fails, all control is lost.

Bus-star layout is commonly used in PLC networks with distributed I/O modules. In a paper mill, for example, PLCs and I/O modules controlling various processes are typically laid out in a linear bus. However, the I/O drops that control chemical kitchens in the pulp preparation area have control points that radiate out in star configurations. If the bus cable is damaged, the star nodes can continue to control local pulp operations. And if a bus node fails, the other star-connected nodes can continue to communicate among themselves.

Star-bus hybrid configuration is often used in spread-out operations such as oil production fields or far-spread petrochemical processing operations. For example, each device, such as an individual oil well, local storage tank, or pump and valve controller, is regulated by nodes on a local bus. The many buses are linked back to a central control room in a star layout. If a local bus cable fails, other local buses in the layout continue to control their respective processes.

Ring-bus and ring-star configurations show how a ring network can be combined with either a bus or a star network. One node on the ring can be one of several nodes on a local bus, or it can be the central node for a local star. The ring is connected in a self-healing configuration for the most critical elements of the network, with less critical nodes connected in bus or star.

Quadruple-hybrid combinations combine all four basic networks into a ring-bus-star collapsed backbone configuration. The most critical items are connected in a self-healing ring; other items are connected in either bus or star or collapsed backbone as best satisfies their location and application.

Redundant Systems

While a hybrid layout can improve performance and enhance reliability, ultimate failure resistance often requires a redundant system–that is, a second or duplicate system that takes over in the event the first stops functioning. Depending on the layout and hazards involved, many system designers opt to duplicate the cable only. The redundant systems, however, need not be identical. The primary system might be fiber optic, with the backup system electrical, and wired with copper conductors. For maximum reliability in bus and star layouts, cables of the redundant systems should be placed in separate cableways some distance from the primary systems. Setting up a second cableway can be enormously expensive.

Duplicating only the cable poses a significant disadvantage in systems with long distances between modems. In order to send the signal down the duplicate cables, each modem needs an optical splitter and combiner. Each splitter introduces a signal loss of 3 to 6 dB, which can create distortion unless the distance between modems is limited.

In a star layout, if the central node is especially at risk of destruction, one option is to duplicate only the star node. It is usually not necessary to duplicate the outlying nodes, because other portions of the system can continue to operate if one node fails. Even duplicating only the star node can be expensive. This is because the central node is very powerful or complex. As with duplicate cables, the duplicate central node should be located some distance from the first, which also increases costs. Thus, for most industrial applications where operations must be preserved in the face of anticipated cable damage, a ring network cabled in a self-healing configuration usually provides the reliability needed at the most moderate cost.

The next part of chapter provides guidelines for fiber optic design considerations with regard to system migration. It covers the following areas: 10BaseF connection to FDDI, token ring connection to FDDI, and FDDI connection to ATM.

System Migration: Moving to Future Networks

The size of your networks and the distance between connections on your networks will depend on the type of signal, the signal speed, and the transmission media (the type of cable used to transmit the signals). For example, the most commonly used fiber optic medium type is the link segment. There are two fiber optic link segments in use, the original Fiber Optic Inter-Repeater Link (FOIRL) segment, and the newer 10BaseFL segment.

The original FOIRL specification from the Ethernet standard of the early 1980s provided a link segment of up to 1000 meters between two repeaters only. As the cost of repeaters dropped and more and more multiport repeater hubs were used, it became cost-effective to link individual computers to a fiber optic port on a repeater hub. Vendors created outboard FOIRL MAUs to allow this, although a repeater-to-DTE (Data Terminal Equipment) fiber connection was not specifically described in the FOIRL standard.

The distance and rate limits in these descriptions are the IEEE-recommended maximum speeds and distances for signaling. For instance, the recommended maximum rate for V.35 is 2 Mbps, but it is commonly used at 4 Mbps without any problems.

Note:

Even though you can usually get good results at speeds and distances far greater than those listed in this part of the chapter, exceeding the maximum distances is not recommended or supported. If you understand the electrical problems that might arise and can compensate for them, you can get good results with rates and distances greater than those shown in this chapter; however, do so at your own risk.

10BaseF Connection to FDDI

To deal with connections to FDDI and other aspects of fiber optic Ethernet, a set of fiber optic media standards, called 10BaseF, was developed. This set of fiber standards includes revised specifications for a fiber optic link segment that allow direct attachments to computers. The full set of 10BaseF specifications includes three segment types:

10BaseFL

The 10BaseFL standard replaces the older FOIRL specifications and is designed to interoperate with existing FOIRL-based equipment. 10BaseFL provides a fiber optic link segment that may be up to 2000 meters long, providing that only 10BaseFL equipment is used in the segment. If 10BaseFL equipment is mixed with FOIRL equipment, then the maximum segment length may be 1000 meters.

A 10BaseFL segment may be attached between two computers, or two repeaters, or between a computer and a repeater port. Because of the widespread use of fiber links, 10BaseFL is the most widely used portion of the 10BaseF fiber optic specifications, and equipment is available from a large number of vendors.

10BaseFB

The 10BaseFB specifications describe a synchronous signaling backbone segment that allows the limit on the number of repeaters that may be used in a given 10 Mbps Ethernet system to be exceeded. In other words, the 10BaseFB specification is a synchronous Ethernet link between repeaters that extends the limit for repeaters and segments in a single unbridged network.

10BaseFB links typically attach to repeater hubs, and are used to link special 10BaseFB synchronous signaling repeater hubs together in a repeated backbone system that can span long distances. Individual 10BaseFB links may be up to 2000 meters in length. This system has a limited market and equipment is available from only a few vendors.

10BaseFP

The Fiber Passive system provides a set of specifications for a fiber optic mixing segment that links multiple computers on a fiber optic media system without using repeaters. In other words, the 10BaseFP specification is a passive star configuration for fiber optics. The signal is shared with other fiber arms using a optical distribution system.

10BaseFP segments may be up to 500 meters long, and a single 10BaseFP fiber optic passive star coupler may link up to 33 computers. This system has not been widely adopted and equipment does not appear to be generally available.

FDDI Connections

The distance limitations for singlemode and multimode Fiber Distributed Data Interface (FDDI) stations are listed in Table 10—1 [2].Table 10—2 summarizes the characteristics of IEEE 802.3 Ethernet and Ethernet 10BaseFL [Cisco, 10]. The distance limitations for 10 Mbps transmission over multimode optical fiber cables are shown in Table 10—3. Table 10—4 lists multimode optical fiber parameters required for 10BaseFL.

Table 10—1: FDDI maximum transmission distances.

Transceiver Type Maximum Distance Between Stations

Singlemode 6.2 miles (10 km)1

Up to 9.3 miles (up to 15 km)2

Multimode Up to 1.2 miles (2 km)2

1 For AGS+( modular router with slots for nine cards) applications of FDDI.
2 For the VIP (Versatile Interface Processor) singlemode FDDI port adapter using SC-type
optical fiber.

 

Table 10—2: IEEE 802.3 Ethernet and Ethernet 10BaseFL physical characteristics.

Parameter IEEE 802.3 Ethernet 10BaseFL Ethernet

Data rate 10 Mbps 10 Mbps

Signaling method Baseband Baseband

Media 50 ohm coax (thick) Multimode optical fiber

Topology Bus Star

 

Table 10—3: 10 Mbps 10BaseFL transmission cable distance limitations

Parameter ST Connections

Maximum segment 1,322 ft (407 m) for any repeater-to-DTE fiber
lengths segment. 1,650 ft (508 m) with five repeaters
and six segments. 3,290 ft (1012 m) for any
inter-repeater fiber segment. 6,571 ft (2022 m)
without a repeater.

Cable specification Multimode fiber optic cable1

Commercially available cables.

 

Table 10—4: 10BaseFL multimode optical fiber parameters.

Parameter Multimode

Attenuation < 3.85 dB/km, at 860 nanometers (nm)

Bandwidth > 170 MHzkm, at 860 nm

Insertion loss < 13.5 dB, at 860 nm

Propagation delay < 6 microseconds/km

Size 62.5/125 micrometer (nominal diameter) optical fiber1

1 IEC Publication 793-2 specification.

 

Note: 

The singlemode and multimode optical fiber connections conform to the following optical power parameters: output power: —20 to —15 dBm; input power: —32 to —15 dBm; and input sensitivity: —32 dBm @ 2.6x11-11 BER @ 126 Mbps.

Token Ring Connection to FDDI

Another way to deal with connections to FDDI is through the fiber optic token ring. For example, an FTB (FDDI to token ring Translation Bridge) connects a token ring departmental LAN to an FDDI backbone as shown in Figures 10—1 and 10—2 [3]. The FTB plugs into the backplane of a modular token ring hub and connects to the FDDI interface from its front panel.

Figure 10—1: FDDI to token ring translation bridge.



Figure 10—2: FDDI to token ring translation bridge where the FTB can be
configured as a dual-attached or single-attached FDDI station.



The FTB can segment the network for improved performance and response times on each departmental LAN as well as between LANs. Servers can connect directly to the FDDI backbone and can communicate with token ring clients, providing a cost-effective, high-performance solution to client servers. In addition, FDDI’s dual-ring capability, together with the FTB’s singlemode and multimode FDDI interfaces, promotes a high-performance fault-tolerant backbone that operates over extended physical distances, ideal for campus networks.

The FTB performs positive filtering on the FDDI side and negative filtering on the token ring port side. All addresses on the token ring side are automatically learned by the bridge. An aging process causes inactive stations to be deleted from the bridge tables.

Any number of FTBs can be connected to the FDDI backbone. Additional bridges (transparent and source routing) can be connected on the token ring LANs, supporting up to 256 stations on the token ring side.

The FTB can be configured as a dual-attached or single-attached FDDI station as shown in Figure 10.2. It supports backup (dual homing) connection on the tree, using for example, port B as the main connection and port A as the redundant connection. For greater reliability, ports A and B can be connected to two different concentrators.

The FTB can be configured and fully diagnosed via a management port. Masks by protocol or by MAC ( Media Access Control) address can be entered via the port, as well as by software downloading.

The FTB performs extensive testing and diagnostics. Whenever a problem is detected, a combination of LEDs (light emitting diodes) indicate the nature of the fault. If an FDDI Bypass Switch is installed, the FTB is bypassed.

Configuration and monitoring of the FTB can also be performed via the in-band Simple Network Management Protocol (SNMP) agent. This enables management by a SNMP Network Management System for LAN/WAN networks or any other SNMP management station. Alternatively, configuration and diagnostics can be performed from a terminal connected to a module management port.

The RISC processor architecture ensures a forwarding rate of 15,000 frames per second. The FDDI and token ring frames are filtered by hardware filters which support the maximum frame rates on both FDDI and token ring.

FDDI Connection to ATM

In reverse, a way to deal with connections from instead of to FDDI is through the fiber optic ATM. Here, the market gurus are not perfectly synchronized concerning the migration to ATM. Some of them advocate skipping the current high-speed LAN standard (FDDI) and moving directly to ATM. Some view such a strategy as risky for most corporate users because it could run into severe interoperability problems.

The ATM principles are to establish a completely transparent LAN/WAN network where a single technology could handle different speeds as well as match the needs (and the budgets) of corporate workgroups. The ideal picture of ATM implementation is a real end-to-end connection based on the same model as the telephone system.

However, most pictures of ATM deployment assume that it will be incorporated into the corporate LAN as a collapsed backbone. Standard LAN segments will be attached to this backbone via routers or special adapter interfaces on the ATM backbone switch. However, the needed interface standards between ATM and other subsystems is not stable. And even when they are finalized, first implementations may not be too efficient or may require extreme care to be tuned properly.

For example, Fibronics On-Demand Bandwidth Hierarchy (ODBH) is based on a transparent, then secure, four-layer evolution to ATM for the installed base of LAN protocols and equipment as shown in Figure 10—3 [4]. ODBH includes a series of high-speed interfaces together with the open GigaHUB architecture. In terms of costs, On-Demand Bandwidth Hierarchy provides a very attractive investment scheme that provides the most cost-effective solution to single user work group, and enterprise bandwidth problems.

Figure 10—3: On-Demand Bandwidth Hierarchy (ODBH).



Layer 1

Shared LANs is the first layer as shown in Figure 10—3. It is the most popular networking implementation for office automation, data entry, client/server, etc.

Layer 2

Ethernet switching significantly increases network performance for workgroups or single users. Private LANs consist of two switching methods to match the customer’s environment as shown in Figure 10—3. Switching simply provides multiples of 10 Mbps, allowing low speed data stream applications to gain performance at low cost.

Layer 3

FDDI is the best-suited technology for client/server applications. FDDI provide extensive reliability, 100 Mbps throughput, and copper or fiber support. In addition, FDDI collision-less technology enables the Ethernet LAN to operate with efficiency. Furthermore, FDDI and Ethernet are completely interoperable with all existing FDDI and Ethernet equipment including adapter cards, concentrators, bridges, routers and standard FDDI interfaces.

This level meets the users bandwidth requirements (fast links) at the desktop and at the server level as shown in Figure 10—3. Combining switched Ethernet and 100 Mbps will ensure a smooth network capacity upgrade for the more powerful computing devices.

Layer 4

Migrating to ATM can then be introduced at no risk of a network architecture change nor performance degradation at the server level or at the LAN level. With FDDI becoming more of a front-end LAN, new backbone implementation can be achieved through ATM. The lack of full ATM management today is not a major issue in this configuration. FDDI provides the pipe for local client/server applications while ATM will provide high-speed LAN interconnection (high speed switching), until the technology reaches a price level compatible with the desktop environment as shown in Figure 10—3.

During the process of deciding which optical fiber design to use and which system to migrate to, it is very important to keep your costs down. Next, we’ll look at how you can maintain a lower optical loss budget.

Optical Loss Budget

The faster design and drafting can be completed, the lower the design costs will be. The more consistent and accurate the calculations and drafting are, the more efficient field construction and operations become.

For example, the typical fiber optic cable used for a fiber link segment is a multimode fiber cable (MMF) with a 62.5 micron fiber optic core and 125 micron outer cladding (62.5/125). Each link segment requires two strands of fiber, one to transmit data, and one to receive data. There are many kinds of fiber optic cables available, ranging from simple two-strand jumper cables with a PVC (Permanent Virtual Circuit) outer jacket material on up to large inter-building cables carrying many fibers in a bundle.

The fiber connectors used on link segments are generally known as ST connectors. The formal name of this connector in the ISO/IEC international standards is BFOC/2.5. The ST connector is a spring-loaded bayonet connector whose outer ring locks onto the connection, much like the BNC (Bayone-Neill-Concelman) connector used on 10Base2 segments. The ST connector has a key on an inner sleeve and also an outer bayonet ring. To make a connection, you line up the key on the inner sleeve of the ST plug with a corresponding slot on the ST receptacle, then push the connector in and lock it in place by twisting the outer bayonet ring. This provides a tight connection with precise alignment between the two pieces of fiber optic cable being joined.

Note: 

There are two other dominant types of optical connectors besides ST, SMA and SC. The SC connector is new, but is or has become the standard.

The wavelength of light used on a fiber link segment is 850 nanometers (850 nm). The optical loss budget for a fiber link segment must be no greater than 12.5 dB. The loss budget refers to the amount of optical power lost through the attenuation of the fiber optic cable, and the inevitable small losses that occur at each fiber connector.

The more connectors you have and the longer your fiber link cable is, the higher the optical loss will be. Optical loss is measured with fiber optic test instruments that can tell you exactly how much optical loss there may be on a given segment at a given wavelength of light. A standard grade fiber optic cable operating at 850 nm will have something in the neighborhood of 4 dB to 5 dB loss per 1000 meters. You can also expect something in the neighborhood of 0.5 to around 2.0 dB loss per connection point, depending on how well the connection has been made. If your connectors or fiber splices are poorly made, or if there is finger oil or dust on the connector ends, then you can have higher optical loss on the segment.

The older FOIRL segment typically used the same type of fiber optic cable and connectors and had the same optical loss budget. The 10BaseFL specifications were designed to allow backward compatibility with existing FOIRL segments. The major difference is that the 10BaseFL segment may be up to 2,000 meters in length if only 10BaseFL equipment is used on the segment.

Fiber on the backbone, copper to the desktop. For years, that’s been the book on premises wiring management. That book has now been rewritten due to the dramatic cut in cost of installing fiber to the desktop. The continued price reductions for fiber cabling and components have brought the overall cost of fiber installation close to that of Category 5 unshielded twisted pair (UTP) copper wiring. This has resulted in greater business profitability by bringing fiber to the desk–the final topic of discussion for this chapter.

Fiber-to-the-Desk

The time for considering optical fiber as the main cabling medium for building cabling has finally arrived. No longer should fiber optics be considered an alternative to copper used only for applications with special requirements. Fiber optic technology is clearly superior in performance and is now competitive in price with the high-end twisted-pair cable required for today’s high-speed networks.

Twisted-pair cable is the most prevalent type of cable used in wiring new buildings. These cables come in several grades based on performance: Category 3 for applications to 16 MHz, Category 4 for applications to 20 MHz, and Category 5 for applications to 100 MHz. Each supports cable runs of up to 100 meters. A standard fiber optic cable for building use can handle applications of several hundred megahertz at distances in excess of 2000 meters. As will be discussed in this chapter, Category 5 cable, the cable required for emerging high-speed applications like asynchronous transfer mode (ATM) and 100 Mbps Ethernet, present some challenges in installation and operation.

The bottom line in the fiber-versus-copper debate is this: fiber optic cable has a performance edge. Copper, on the other hand, is a more widely understood and accepted technology. More important, the costs of fiber components are competitive with their copper counterparts. And, if you add life cycle costs, including the costs of downtime and possible obsolescence, fiber is the better value.

The Seven Advantages of Fiber

Fiber optics would not even be considered if it did not offer distinct advantages over traditional copper media. These advantages translate into the following:

• Information-carrying capacity.

• Low loss.

• Electromagnetic immunity.

• Light weight.

• Smaller size.

• Safety.

• Security.

Information-Carrying Capacity

Fiber optic cable offers bandwidth well in excess of that required for today’s network applications. The 62.5/125-micrometer fiber recommended for building use has a minimum bandwidth of 160 MHz-km (at a wavelength of 850 nm) or 500 MHz-km (at 1300 nm). Because bandwidth is a product of frequency and distance, the bandwidth at 100 meters is over 1 GHz. In comparison, Category 5 cable is specified only to 100 MHz over the same 100 meters.

With the high-performance singlemode cable used by the telephone industry for long distance telecommunications, the bandwidth is essentially infinite. That is, the information-carrying capacity of the fiber optic cable far exceeds the ability of today’s electronics to exploit it as shown in Figure 10—4 [5].

Figure 10—4: The bandwidth of optical fibers comfortably surpasses the needs of today’s applications and gives room for growth.



Low Loss

A fiber optic cable offers low power loss. Low loss permits longer transmission distances. Again, the comparison with copper is important–in a network, the longest recommended copper distance is 100 meters; with fiber optic, it is 2000 meters as shown in Table 10—5 [AMP, 3].

Table 10—5: Copper Versus Fiber.

Copper    Fiber
Multimode Singlemode

Bandwidth (100 meters) 100 MHz 1 GHz > 100 GHz

Transmission distance 100 meters 2000 meters 40,000 meters

FCC EMI concerns Yes              No

EMI susceptibility Yes              No

Crosstalk Yes              No

Ground loop potential Yes              No

Weight Heavier            Lighter

Size Larger            Smaller

A principal drawback of copper cable is that loss increases with the signal frequency. This means high data rates tend to increase power loss and decrease practical transmission distances. With fiber, loss does not change with the signal frequency.

Electromagnetic Immunity

By some estimates, 60% of all copper-based network outages are caused by cabling and cabling-related products. Crosstalk, impedance mismatches, and EMI susceptibility are major factors in noise and errors in copper systems. What’s more, such problems can increase with incorrectly installed Category 5 cable, which is more sensitive to poor installation than other twisted-pair cable.

Because a fiber optic cable is a dielectric, it is immune to electromagnetic interference. It does not cause crosstalk, which is a critical limiting factor for twisted-pair cable. What’s more, it can be run in electrically noisy environments, such as a factory floor, without concern because electrical noise will not affect fiber. There’s no concern with proximity to noise sources like power lines or fluorescent lights. In short, fiber is inherently more reliable than copper.

Light Weight

Fiber optic cable weighs less than comparable copper cable. A dual-fiber cable is 20% to 50% lighter than a comparable 4-pair Category 5 cable. Lighter weight makes fiber easier to install.

Smaller Size

Fiber optic cable has a smaller cross section than the copper cables it replaces. Again, relative to Category 5 twisted-pair cable, a duplex optical fiber takes up about 15% less space.

Safety

Since the fiber is a dielectric, it does not present a spark hazard. What’s more, cables are available with the same flammability ratings as copper counterparts to meet code requirements in buildings.

Security

Fiber optic cable is quite difficult to tap. Since it does not radiate electromagnetic energy, emissions cannot be intercepted. And physically tapping the fiber takes great skill to do undetected. Thus, the fiber is the most secure medium available for carrying sensitive data.

Basically, fiber optic cable offers high bandwidth over greater distances with no danger of electrical interference. Its small size and lighter weight give it an installation edge for pulling and installing, especially in tight spaces. And it’s safe and secure. The clear advantages of fiber optics are too often obscured by concerns that may have been valid during the pioneering days of fiber, but that have since been answered by technical advances.

The Four Myths about Fiber Optics

A great deal of discussion has focused on the capabilities and shortcomings of high performance copper and fiber optic cables for the horizontal wiring system. Many misconceptions have clouded the facts surrounding the capabilities, survivability, and craft-friendliness of fiber. Technology improvements (coupled with more cost-effective solutions) have made fiber a viable and valuable option for high-performance wiring to the workstation. This part of the chapter dispels four myths which have surfaced about the feasibility of fiber as a complete communications system solution–from backbone links all the way to the desk.

Myth #1: Fiber Optic Cable is Fragile

An optical fiber has greater tensile strength than copper or steel fibers of the same diameter. It is flexible as shown in Figure 10—5 [AMP, 3], bends easily, , and resists most corrosive elements that attack copper cable. Optical cables can withstand pulling forces of more than 150 pounds–about six times that recommended for Category 5 cable. Fact is, Category 5 cable may be more fragile than optical cables: tight cable ties, excessive untwisting at the connector, and sharp bends can degrade the cable’s performance until it no longer meets Category 5 performance requirements.

Figure 10—5: Fiber optic cable is flexible, sturdy and easy to work with.



Myth #2: Fiber Optic Cable Is Hard to Work With

Myth 2 derives from the early days of fiber optic connectors. Early connectors were difficult to apply; they came with many small parts that could tax even the nimble fingered. They needed epoxy, curing, cleaving, and polishing. On top of that, the technologies of epoxy, curing, cleaving, and polishing were still evolving. Today, connectors have fewer parts, the procedures for termination are well-understood, and the craftsperson is aided by polishing machines and curing ovens to make the job faster and easier.

Even better, epoxyless connectors eliminate the need for the messy and time-consuming application of epoxy. Polishing is an increasingly simple, straightforward process. Pre-terminated cable assemblies also speed installation and reduce a once (but no longer) labor-intensive process. Also, pre-engineered parts, cables, connectors, etc. make using fiber optics much easier.

Myth #3: Fiber Is Expensive

Fiber optic cable and related components are comparably priced to Category 5 copper counterparts. As fiber optic cable becomes increasingly easier to work with, it means that installation costs are becoming less expensive. Pulling costs are the same. Termination time is about equal (Category 5 cable takes more care, and time, than other UTP).

Is fiber more expensive than copper? On the installed cost side, fiber optic cable and Category 5 components are comparably priced. On the life cycle costs, fiber may be cheaper in the long run. Many users have found fiber optic cable installations easier to maintain and more reliable. The cost of network downtime and glitches can be enormous: the cable plant that minimizes these costs justifies the small premium in components/installation costs.

What’s more, costs are changing every day and for every installation. Don’t assume, any longer, that fiber optics cost more.

Myth #4: Fiber Has No Place to the Desktop

How much bandwidth is needed at the desktop? How much is too much? Underestimating a user’s requirements has long been a mistake in the computer and networking industry. Critics of fiber optics argue that users don’t need high bandwidth to the desktop, so that using fiber is wasted potential. Data rates of 25 Mbps and 51 Mbps are among those touted for desktop connectivity.

Yet, 4 Mbps token ring seemed fast at first, only to be replaced by the four-times-faster 16 Mbps version. Traditional Ethernet is being replaced by 100 Mbps flavors. Video, video teleconferencing, multimedia, and other bandwidth-hungry applications are vying for bandwidth. While the network backbone can certainly benefit from the high-speed, long-distance transmission achieved with fiber optics, don’t count the desktop out. Demands at the desktop will invariably grow as new applications like ATM require more bandwidth.

Technology presents an important issue here. Category 5 cable may have been pushed to its limits in achieving 100 MHz performance. While advances in data encoding schemes may provide higher data rates within this frequency framework, the question remains whether Category 5 performance can be extended. Will we need Category 6 UTP? At the same time, fiber optic technology has room to spare. For the 100-meter distances recommended for links to the desktop, Category 5 cable has a bandwidth of 100 MHz, while the fiber optic bandwidth is over 1 GHz.

Today, networks can be the lifeblood of businesses, universities, and medical centers, so it is essential that the cable system perform reliably to the specifications required. Downtime can be costly. In the final analysis, fiber for workstation applications offers long-term performance, security and reliability advantages, and should be considered as part of the network solution into the 21st century.

From Here

This chapter presented an overview of fiber optic design considerations (layout, system migration, loss budget calculations, and fiber to disk, etc.). The next chapter discusses wireless design considerations (spread spectrum, microwave, infrared, wireless WANs and LANs, etc.).

End Notes

[1] John C. Huber, Ph.D., PE, "Understanding Fiber Optics: Selecting the Optical
Network," Fiber Optics Laboratory Manager, 3M Telecom Systems Group, Austin, TX, 9May 8, 1995), p. 1.

[2] "Site Preparation," (Some material in this book has been reproduced by Prentice Hall with the permission of Cisco Systems Inc.), COPYRIGHT © 1998 Cisco Systems, Inc., ALL RIGHTS RESERVED, 170 West Tasman Drive, San Jose, CA 95134-1706, USA, 1997, pp. 9—10.

[3] "RADring Module: FTB-FDDI to Token Ring Translation Bridge," RAD Data Communications, 900 Corporate Drive, Mahwah, NJ 07430, and Corporation UNI, Russia,
1998, p. 4.

[4] "Product Strategy," NBASE Communications, Fibronics International, 16 Esquire Road, North Billerica, MA 01862, 1996, p. 1.

[5] "Why Fiber? Why Now," Reprinted with the permission of AMP Incorporated, Investor Relations, 176-42, PO Box 3608, Harrisburg, PA USA 17105-3608, 1997, p. 2.

Page 1 | Next Chapter

Vendor Comparisons
Network Computing’s Vendor Comparisons provide extensive details on products and services, including downloadable feature matrices. Our categories include:

Research and Reports

Network Computing: April 2013



TechWeb Careers