Computer Networks Prof.
Hema A Murthy
Fibre Distributed Data Interface
• Runs on fibre and not copper
• dual ring
– two independent rings transmitting data in
opposite direction
– second not used for normal operation
– used only if primary fails
Indian Institute of Technology Madras
Computer Networks Prof. Hema A Murthy
FDDI Ring
Indian Institute of Technology Madras
Computer Networks Prof. Hema A Murthy
FDDI Ring
• Expensive – twice the amount of fibre
– stations may be allowed to connect on a single
cable
• single attachment station (SAS)
• use concentrator to connect several SASs to
dual ring
Indian Institute of Technology Madras
Computer Networks Prof. Hema A Murthy
Fibre Concentrator
Upstream Downstream
neighbour neighbour
SAS SAS SAS SAS
Concentrator detects failure of SAS
- Optical bypass to isolated failed SAS
Indian Institute of Technology Madras
Computer Networks Prof. Hema A Murthy
FDDI Ring
• Each NE Adapter hosts some number of bits
between its input and output interfaces
– Variable buffer size
• 9 ≤ buffersize ≤ 80 bit
• Station transmits an amount equal to half
buffer
• Total time depends on buffer
Indian Institute of Technology Madras
Computer Networks Prof. Hema A Murthy
Delay in FDDI
• Example: 100 Mbps FDDI
• - 10 ns for bit time
• - Each station 10 bit buffer – waits until
buffer half full before transmitting
– station introduces 50 ns delay into TRT
Indian Institute of Technology Madras
Computer Networks Prof. Hema A Murthy
FDDI –Physical Characteristics
• 500 stations with a maximum distance of
2km between any pair
• maximum network length : 200km
• 100 km connecting all stations (dual ring)
Indian Institute of Technology Madras
Computer Networks Prof. Hema A Murthy
FDDI –Physical Characteristics
• FDDI encoding:
– 4B/5B encoding
– Replace 4B with 5B code such that no more
than one leading zero,
– no more than two trailing zeros and no more
than 3 consecutive zeros
Indian Institute of Technology Madras
Computer Networks Prof. Hema A Murthy
Asynchronous vs. Synchronous
Traffic
• Synchronous traffic
– Traffic is delay sensitive
– station transmits data whether token is late or
early
– But synchronous cannot exceed one TTRT in
one TRT
• Asynchronous traffic
– Station transmits only if token is early
Indian Institute of Technology Madras
Computer Networks Prof. Hema A Murthy
Measurement of Token Rotation
Time (TRT)
• Target Token Rotation Time (TTRT – agreed
upon time)
• Time between successive token arrival – TRT
observed by any node
• TRT > TTRT
– token late station does not transmit data
• TRT < TTRT
– station holds token until TTRT
– down stream station may not be able to transmit
Indian Institute of Technology Madras
Computer Networks Prof. Hema A Murthy
Token Maintenance
• Process of setting up TTRT
• Monitor ring to ensure token has not been lost
• Fix TTRT – each node bids for the TTRT
• Idle time between valid transmissions that a given
node experiences is
– ring latency + time to transmit a full frame
– 2.5 ms maximally sized ring
• If timer expired then claim token
– TTRT lower used
– Lower TTRT – new node enters the bidding process by
Indian Institute of Technology Madras
Computer Networks Prof. Hema A Murthy
FDDI: Analysis
• Worst Case
– Nodes with asynchronous traffic use one TTRT
– Next nodes with synchronous traffic in one
TTRT
• TRT at a node = 2 * TTRT
– Synchronous traffic TTRT
– Next no asynchronous – token late
Indian Institute of Technology Madras
Computer Networks Prof. Hema A Murthy
FDDI Analysis
– No back to back transmission of TTRT
• When does a node transmit asynchronous data
– TRT + ε = TTRT => Transmit
– Total TRT = TTRT + full FDDI frame
– if claim frame makes it all the way back to the
original sende
• node knows it is only active bidder => safely claim
the token
Indian Institute of Technology Madras
Computer Networks Prof. Hema A Murthy
FDDI Frame Format
24
Destination Source End of Frame
control CRC
address address frame Status
48 48 32
8
Start of frame
Indian Institute of Technology Madras
