Shining a Light on FDDI

topology design. Fiber-optic cable also offers excellent

Introduction noise immunity, and is virtually impossible to tap.
FDDI - the 100-Mbit/s Fiber Distributed Data Interface
FDDI is a token-passing technology that uses a timed-token
networking technology is a solution for many of the new
protocol to guarantee network access between network
problems presented by changing corporate networks:
stations (network devices and end nodes). Figure 1 shows a
standard dual-attach ring. Network access is negotiated
• Groups that previously had no need for communication
between stations at initialization, and every time a new
now want network connections.
node is added to the network.
• Existing token ring and Figure 1 The network backbone is
Ethernet backbones that
constructed of stations
are interconnected are
interconnected by two
now reaching their
counter-rotating rings. These
capacity.
rings are two pairs of fiber-
optic cable to which each
• Applications require
device is attached. Cable
increasing bandwidth,
lengths between stations can
security, and fault
be anywhere from 2
tolerance.
kilometers (km) with
multimode fiber, to 60 km
These are examples of some
with single-mode fiber. The
of the network requirements
total ring length cannot
that are met by FDDI. At
exceed a maximum of 200
100 Mbit/s, FDDI allows
km. During normal
high-speed interconnection of
operation, the first ring is the
all the LANs in an
primary data carrier, and the
organization's network.
second acts as the backup.
FDDI easily overcomes the
This offers the network a
performance limitations of
greater degree of redundancy
10-Mbit/s Ethernet and 16- FDDI Ring and fault tolerance.
Mbit/s token ring networks. In addition, FDDI provides the
security of fiber-optic cabling as well as fault tolerance that
is built into the network design. As a result FDDI is FDDI and the OSI Model
becoming widely accepted, especially as a network
backbone technology. The FDDI standard is made up of four distinct parts: (1) the
Physical Layer Medium Dependent (PMD) and Single
Mode Fiber Physical Layer Medium Dependent (SMF-
Technology Overview PMD), (2) the Physical Layer Protocol (PHY), (3) the
Media Access Control (MAC), (4) the Station Management
Fiber Distributed Data Interface (FDDI) is an ANSI and (SMT).
ISO specification (X3T9) for the transmission of data at
high speeds, typically 100 Mbit/s, using fiber-optic cable as Although not part of the FDDI standard, the Logical Link
the transmission medium. Optical fiber technology offers Control (LLC) is required by FDDI to assure transmission
networks a great degree of flexibility in bandwidth and

July 1993 1993 Hewlett-Packard Company

Figure 2
SMT frames carry data and control information for the
operation and management of the FDDI network.

The MAC, PHY, and PMD standards were approved by

ANSI and ISO by 1990. The SMT standard was approved
in 1991. FDDI is the only LAN with extensive
management capabilities as defined in the SMT.

How FDDI Works

FDDI is based on two counter-rotating 100-Mbit/s fiber-
optic token-passing rings. If one ring should fail, FDDI
automatically becomes a single, not dual, FDDI ring. The
rings consist of point-to-point connections between
adjacent stations. Stations negotiate for ring access at
initialization and when new stations are added to the ring.
The two rings act as a primary carrier and a backup
(secondary) carrier. Data flows in opposite directions on
FDDI Protocol Stack the two rings.

of user data. These standards define the 100-Mbit/s fiber- FDDI Stations
optic dual counter-rotating FDDI ring.
There are three types of FDDI stations: dual-attach stations
(DAS), single-attach stations (SAS), and concentrators.
The PMD, SMF-PMD, and PHY are equivalent to the
physical layer of the OSI model (see figure 2). The PMD Dual attach stations are devices that connect to both the
and SMF-PMD correspond to the lower portion of the primary ring and the secondary ring simultaneously. Single
physical layer. The PMD defines the media requirements attach stations are devices that are connected through a
for multimode fiber, such as fiber-optic cable, connectors, primary ring only, and are usually attached using
and driver receiver operation for FDDI stations. Cables, concentrators.
connectors and receivers are discussed below in the
“Cabling” section. The SMF-PMD defines similar Dual attach stations have two ports, one labeled port A, the
requirements for single-mode fiber-optic media. The PHY other port B. With each port there is a primary and a
corresponds to the upper portion of the physical layer and secondary ring connection; port A is primary IN secondary
defines the symbol set, link states, encoding/decoding,
clocking, and framing. Figure 3

The MAC corresponds to part of the data-link layer of the

OSI model. The MAC standard defines the token-passing
method as the means for acquiring access to the ring. It is
responsible for frame and token construction, sending and
receiving frames on the FDDI ring, and delivering LLC
frames.

The LLC service provides the transmission of a frame of

data between two stations. LLC frames carry user
information to stations on the ring and also to the extended
LAN.

FDDI also has a process that defines protocols for

managing the PMD and SMF-PMD, PHY, and MAC,
called Station Management (SMT). SMT defines facilities
for connection management, station configuration, error
recovery, and the encoding of SMT frames. MAC and
Dua-Attach Stations on an FDDI Network

2 HP Router Application Note

OUT, port B is primary OUT secondary IN. If either port
fails, the other port then becomes primary IN and OUT on Access to the FDDI network is controlled by a token that
the same port. circulates the primary ring. When a station has data to
transmit, it must capture the token before it can transmit
Concentrators connect directly to the backbone of the ring, data onto the ring. Only one token may exist on the ring at
and provide indirect access to the ring for other devices. any one time. If the token is lost, the ring will enter a phase
Concentrators support multiple DAS and SAS connections, known as beaconing (a state in which the beaconing station
and add a degree of fault tolerance to the network by controls all access to the ring). If more than one token has
isolating end nodes from the ring. been generated, the ring is then scrubbed (scrubbing is used
to remove all data from the ring).
Ring Access Ring Wrap

When a station is first attached to the FDDI ring, it

undergoes an initialization process. During this Ring wrap is one of the three techniques that FDDI uses in
initialization process, a newly inserted station initiates a order to ensure fault tolerance. If any station on the dual
connection management process (CMT) with its neighbor. ring fails or if the cable is damaged, the dual ring
CMT tests port types, performs physical port tests or link automatically wraps into a single ring as shown in Figure 4.
confidence tests, and transmits link-state characters upon
completion of the tests. Stations initiate CMTs between Multiple failures within a single dual ring network present
each upstream and downstream neighbor. other challenges. The likelihood of multiple failures
greatly increases as the number of nodes within an FDDI
Upon successful completion of the initialization phase, the network increases. If a single failure occurs, the ring wraps
ring begins the claim process to generate a token. The and becomes a single ring. If another failure should occur,
claim process begins with all stations negotiating a target the ring is segmented and separated into two rings.
token rotational timer (known as TTRT). Each station
makes a bid for the timer, and the one with the lowest value Optical Bypass
wins the right to initialize the ring by inserting the token. If
two stations happen to make the same bid, the station with Optical bypass can be used for fault tolerance to prevent
the highest address wins the right to generate the token. ring segmentation. Optical bypass switches maintain
Once the claim process has finished, the ring enters a steady connectivity of the FDDI ring in the absence of power or
state, to be disrupted only by the insertion of another during fault conditions in a station. Stations bypassed by
station, or by a break in the ring. optical bypass switches are effectively removed from the
ring. Figure 5 shows an inoperative station switched out by
an optical bypass switch.
Figure 4

Ring Wrap

Shining a Light on FDDI 3

Optical bypass switches are recommended when designing the DAS automatically enables the backup link to the
a multistation FDDI backbone. In a fault condition, a DAS secondary concentrator.
will cause the ring to wrap over the secondary ring,
isolating the bad section of the ring. If this happens to Cabling
more than any single section of the FDDI ring, the network
is sectioned into two independent rings. Stations that have Optical fiber has many advantages over traditional copper
an optical bypass device are physically switched out when a cabling. Fiber doesn't emit electrical signals and is immune
fault condition occurs. Note that they do not prevent the to electrical interference, making it secure and reliable. It is
network from segmenting in a cable fault condition. easy to manipulate because it is lightweight. Important
specifications of fiber-optic cable are the optical signal
There are several limitations of optical bypass that a wavelength and attenuation.
network designer must be aware of:
Two types of transmission media are currently defined by
• When the bypass switch bypasses a station, the the FDDI standards: multimode fiber and single-mode
station is effectively removed, possibly allowing fiber. Both operate at 100 Mbit/s. Multimode means that
the maximum segment length to exceed 2 km. multiple rays of light can enter the fiber from different
angles. Multimode uses light-emitting diodes (LEDs) that
• The integrity of the ring is only as good as the convert electrical signals into light and transmit the light
mechanical integrity of the bypass switch. into the fiber-cable.

• Optical bypass switches can produce up to 2.5 dB Single-mode means that only one ray of light is allowed to
optical power loss. This loss must be considered enter the fiber. Single-mode uses laser diodes (LDs) to
when calculating attenuation. Attenuation and convert electrical signals into light and transmit the light
optical budget are discussed later. into the fiber-optic cable.

Dual Homing Attenuation and Optical Budget

Dual homing is a fault tolerance technique used for
redundancy in concentrator environments where continuous Figure 5
uptime is crucial. Dual-homing allows a DAS to use one
link as a backup link for redundancy purposes. Thus, one
of the two attachments is active at any given time. Figure
6 shows a dual-homed device attached to concentrators.

A station can be dual-homed to the same concentrator or to

two different concentrators. When it is homed to two
different concentrators and the primary concentrator fails,

Figure 6

DAS with Optical Bypass Switch

Signal attenuation through optical fiber is important with

FDDI. It describes the amount of energy (optical power)
that is lost as the light signal travels from the transmitter
Dual Homed Station through the cable to the receiver. The longer the cable, the

4 HP Router Application Note

higher the loss of optical power. Energy loss is denoted in into light signals. In single-mode fiber cable, transmitters
decibels (dB), an expression used to mathematically are laser-quality light-emitting diodes or laser diodes (LDs).
compare the power of two signals.

Attenuation is calculated by knowing the unit attenuation

Internetworking with HP Routers
and the length of the link. The maximum cable attenuation HP currently offers two routers that provide FDDI links:
is the power loss in the cable as well as any loss incurred by the HP 27270B Router CR, and the HP 27290A Router BR.
splices, connectors, and anything else connected to the Both FDDI link modules are compliant with the following
cable. PMD specifies an optical power budget between any ANSI X3T9.5 standards:
two stations of 11 dB. SMF-PMD allows for a range of
power budgets that extends from a minimum of 10 dB to a • Physical Medium Dependent (PMD)
maximum of 32 dB. • Physical Layer Protocol (PHY)
• Media Access Control (MAC)
To calculate the cable loss between two adjoining stations, • System Management (SMT)
use the formula shown below:
The FDDI modules also comply with RFC 1188, which
Attenuation = (Cable Len (km) x Attenuation/km) + specifies transmission of IP datagrams over FDDI media,
(Splices x Attenuation per Splice) + and with IEEE 802.1 Parts D & H.
(Connectors x Connector Attenuation) +
Physically, the FDDI modules provide a single MAC
(Loss at Transmit MIC)
connection and support both Class A and Class B
attachments to FDDI media. A Class A attachment requires
For example:
two physical connectors and provides connectivity between
2 km fiber x 2.5 dB/km 5.0 dB
a dual-attachment station (DAS) and the FDDI primary and
2 splices x 0.25 dB/splice 0.5 dB
secondary rings. A Class B attachment uses a single
1 connector x 0.5 dB/connector 0.5 dB
physical connector and provides connectivity between a
Loss at transmit MIC 0.5 dB
single-attachment station (SAS) and the FDDI primary ring
Total link attenuation 6.5 dB
or an FDDI concentrator. The HP FDDI link interface only
Once you know the attenuation, you can calculate the supports multimode 50-micron or 62.5-micron graded-
remaining power available for the network. The maximum index fiber-optic cable.
allowed loss is 11 dB. Following is the formula: FDDI Routing
Power Available = Optical Power Budget - Attenuation Since routers are media-independent, they encapsulate
For example: packets to conform to the media type in use. RFC 1188
specifies 802.2 LLC with SNAP (subnetwork access
Optical Power Budget 11 dB protocol) when routing TCP/IP traffic over FDDI media.
Attenuation 6.5 dB RFC 1188 conforming routers provide IP routing between
Total Available 4.5 dB source and destination end systems on Ethernet or 802.3
After calculating the dB loss of a particular link, it is
important to compare the resulting power with receiver Figure 7
sensitivity. If the receiver is not sensitive enough, poor link
quality will result.

Optical Transmitters and Receivers

Transmitters convert data from electrical signals to light.
The receiver converts the light signals back to electrical
signals. Receivers contain photo detectors that convert
incoming optical signals back into electrical signals.
Optical transmitters convert modulated electrical signals
into modulated light signals that are transmitted through the
fiber-optic cable. In multimode fiber cable, transmitters are
light-emitting diodes (LEDs) that convert electrical signals
Encapsulation Bridging

Shining a Light on FDDI 5

LANs across an FDDI backbone, as well as to destination
systems directly connected to the backbone. Table 2
Protocol Source Translation Destination
Other routed protocols are translated on the FDDI ring Type Rules Type
according to rules defined by RFC 1188 and IEEE 802.1
bridge standards. Adherence to such rules allows XNS Ethernet RFC 1042 Ethernet

interoperability between multiprotocol routers and DECnet Ethernet RFC 1042 Ethernet

translating bridges that may be attached to the same FDDI IV

ring. The protocols supported on FDDI include: Novell Proprietary RFC 1042 Ethernet

Novell Ethernet RFC 1042 Ethernet

• TCP/IP Appletalk LLC + LLC + LLC +

• Novell IPX Phase 2 SNAP SNAP SNAP

• Appletalk Phase 2 Source 802.2 LLC 802.2 LLC 802.2 LLC

• DECnet Phase IV Routing

IP Ethernet RFC 1042 Ethernet

• Xerox XNS
IP LLC + LLC + LLC +

• HP Probe SNAP SNAP SNAP

Protocol Translation and Encapsulation
Encapsulation Bridging requires a translation. When bridging Ethernet or 802.3
There are many protocols that are non-routable and packets to FDDI media, a MAC-layer translation is also
therefore must be bridged. These non-routable protocols required.
must be encapsulated in an FDDI frame in order to be
transported across the ring. Encapsulation is the bridge's The 802.1 standards define how a received packet is
implementation that enables interconnection of similar formatted for transfer across an intervening LAN (802.x or
networks over an FDDI network. Figure 7 shows how non­ FDDI) and presented to a destination LAN. In general,
routable protocols might be encapsulated across an FDDI Ethernet Version 2 frames are encapsulated according to
network. Within such a topology, a bridge encapsulates the rules specified in RFC 1042, and then converted back to
original Ethernet/IEEE 802.3 frame into a new message Ethernet format. IEEE 802.3/802.2 LLC and
type (in this case, an FDDI-specific packet) for travel 802.3/LLC+SNAP remain unchanged when transferred
across the FDDI ring. At the destination, the message is across an intervening LAN. Table 1, summarizes translation
removed from the FDDI packet and transmitted in its rules. Table 2 summarizes protocol-specific translation.
original form.
If an Ethernet frame generated on LAN 1 is destined for
It is important to note that encapsulation is not the preferred LAN 2, the frame is translated as shown in figure 8.
method of transporting data across an FDDI network. In
addition, no standard method of encapsulation exists, Translation by the bridge consists of:
preventing multivendor interoperability. 1. extraction of addressing information from the
Ethernet header
FDDI Translation Bridging 2. incorporation of address information into a newly
generated FDDI MAC header
To ensure multivendor interoperability, the bridge protocol 3. encapsulation of Ethernet data as specified in RFC
should be based on the IEEE 802.1 Spanning Tree Protocol. 1088
Translation is required when bridging between LANs with
different data link layer characteristics. For example, 4. FCS recalculation
forwarding from an Ethernet to an 802.3 end station 5. addition of the FDDI MAC-level trailer

Table 1
Source Network Translation Destination
Encapsulation Rules Network
Encapsulation
Ethernet RFC 1042 Ethernet
(SNAP OUI
00-00-00)
802.3/LLC + FDDI LLC + 802.3/LLC +
6 HPSNAP
Router ApplicationSNAP
Note SNAP
802.3/802.2 LLC FDDI 802.2 802.3/802.2
LLC LLC
Standalone FDDI Rings

Network designers can take advantage of FDDI's high-

speed capabilities without an FDDI backbone in a network
topology. In the absence of a ring, a network designer can
use the HP 27290A Router BR as a standalone FDDI
network, interconnecting a maximum of two SAS devices
or a single DAS.

Figure 9

Standalone FDDI Ring

A file server is a very good candidate for a standalone

FDDI network. In a standard configuration, a server would
be connected directly to an Ethernet segment. Traffic

Figure 8

Translation Bridging
between segments would be very heavy, with traffic on the
server's segment being the heaviest. If servers were
connected to the HP 27290A Router BR as shown in figure

Shining a Light on FDDI 7

9, two benefits would be: 1. The traffic between segments
would decrease drastically, effectively increasing the
bandwidth of the ethernet internet. 2. Each server would
have 100 Mbit/s bandwidth with which to serve all
incoming segments.

In general, a device can be connected to the HP Router BR

creating a dual-attach ring with only two devices, or as a
SAS, creating a single-attach ring.

8 HP Router Application Note