Title: Midterm Exam
1Midterm Exam
- March 29 (Thursday) 2 weeks from now
- April 3 (Tuesday)
2Performance of CSMA/CD
Derivation of maximum throughput of CSMA/CD
- Let A be the probability that some station can
successfully transmit in a slot. We get - In the above formula, A is maximized when P1/ N.
Thus
3Performance of CSMA/CD
- Prob contention interval has a length of j
slots - Prob 1 successful attempt x Prob j-1
unsuccessful attempts
The expected number of slots in a contention
interval is then calculated as
4Performance of CSMA/CD
- Now we can calculate the maximum efficiency of
CSMA/CD with our usual formula
5Performance of CSMA/CD
- If N 100, A 0.9999 0.3697
- Efficiency 1 / (1 5.4a)
- If we have just one computer wanting to transmit,
then we can get almost 100 efficiency. - The above equation is when we have many computers
wanting to transmit at the same time (which is
the common case)
6Performance of CSMA/CD
- Given an Ethernet LAN with a 2.5 km long
- Assume average packet size is 620 bits
(practical) - Propagation delay 2500 / (2 x 108) 1.25 x
10-5 sec. - Transmission time of a packet 620 / (10 x 106)
6.2 x 10-5 sec - a 1.25 x 10-5 sec / 6.2 x 10-5 sec 0.2016
- Efficiency 1 / (1 5.4 x 0.2016) 48
7Performance of CSMA/CD
- In this case, the effective transmission rate of
the 10-Mbps Ethernet is only 4.8 Mbps. - If the Ethernet transmits TCP/IP frames, about 30
bytes are for header. - Hence only (620 240) / 620 61 of the frame
bits is user data bits - Therefore, the maximum rate at which the Ethernet
network can transmit user data is 61 x 4.8 Mbps
2.928 Mbps. - This efficiency is very sensitive to the length
of packets.
8Performance of CSMA/CD
- Suppose we have a 200 m, 1 Gbps shared Ethernet.
- We have the same average packet length of 620
bits. - Propagation delay 200 / (2 x 108) 1.0 x 10-6
sec. - Transmission time of a packet 620 / (1 x 109)
0.62 x 10-6 sec - a 1.0 x 10-6 sec / 0.62 x 10-6 sec 1.613
- Efficiency 1 / (1 5.4 x 1.613) 10.3
- That is why we find only switched Gbps Ethernet
9Average Delay of CSMA/CD
- Recall the efficiency of the Ethernet is
- Ttime / (Ttime 5.4 Propagation delay)
10Average Delay of CSMA/CD
- Before each successful transmission, 5.4 x
propagation delay is wasted - A cycle consists of 5.4 x propagation delay
Ttime - Given N nodes wanting to transmit, we assume that
each node gets to transmit after N cycles - Average time N x (5.4 x propagation delay
Ttime)
11Average Delay of CSMA/CD
- Given an Ethernet LAN with a 2.5 km long
- Assume average packet size is 620 bits
(practical) - Assume, we have 40 computers connected
- Average time N x (5.4 x propagation delay
Ttime) - 40 x (5.4 x 1.25 x 10-5 sec 0.62 x 10-6 sec)
2.72 10-3 sec - This is reasonable (but not guaranteed) for voice
applications (phone conversation) - since this Ethernet can accommodate the bandwidth
needed (40 x 64 kbps 2.5 Mbps). - Unlike Video applications where it would be too
much (40 x 1.5 Mbps 60 Mbps)
12Token Ring Priority Scheme Example
- (a) Station A has a frame with a priority of 4.
Stations B and D have frames of priorities 5 and
6, respectively. - (1) A transmitted the frame (P4,R0)
- (2) B makes a reservation (P4,R5)
- (3) D found that its priority is higher than R
(6gt5) - (4) D also makes a reservation (P4,R6) to
overwrite the old reservation - (b) Station A received the frame
- (1) A issues a free token with a priority of 6
(P6,R0) and keeps in mind that it upgrades the
priority from 4 to 6 by pushing Sr4,Sx6 (stacks
to store the old priority level and the new
priority level, respectively) - (2)B makes a second reservation (P6,R5) on this
token - (3) D gets this token (Pm P 6)
- (4) D transmits a frame with a priority of 6 and
a reservation of 5.
13Token Ring Priority Scheme Example
- (c ) Station C has a frame with a priority of 7
- (1) The frame transmitted by D passing B (P6,
R5) - (2) B bypasses the frame. C found that this is a
data frame and it has a higher priority (PmgtR). - (3) C makes a reservation (P6,R7) to overwrite
the old reservation. D begins to remove the
transmitted frame - (d) Station D received the frame
- (1) D issues a free token with a priority of 7
(P7,R0), and keeps in mind that it upgrades the
priority from 6 to 7 by pushing Sr6,Sx7. - (2) A bypasses this frame
- (3) B makes a reservation again (P7,R5) on this
token - (4) C gets this token and transmits a frame with
(P7,R5).
14Token Ring Priority Scheme Example
- (e) C transmitted the frame.
- (1) The frame transmitted by C passing A
(P7,R5) - (2) B bypasses the frame. C begins to remove this
frame. - (3) C issues a free token with (P7,R5)
- D found that the priority level in the incoming
token (7) is upgraded by itself and downgrades
the priority level from 7 to 6 by popping the Sx
and Sr. - (f) Station A received the downgraded token
- (1) A found that the priority level in the
incoming token (6) is upgraded by itself and
tries to downgraded the priority level from 6 to
4. Nevertheless, it also found that the reserved
priority level in the token is 5. As a result,
the priority level is upgraded to 5 (Sx5,Sr4) - (2) A issues a token with (P5,R0). B finally
found a token for itself. - (3) B uses this token to transmit a frame to C.
15Token Ring Priority Scheme Example
16(No Transcript)
17Performance of Token Rings
- Parameters and Assumption
- End-to-end propagation delay a
- Packet transmission time 1
- Number of stations N
- Assume that each station always has a packet
waiting for transmission - Note The ring is used either for data
transmission or for passing the token
18Performance of Token Rings
- Define
- T1 Average time to transmit a frame. Per
assumption, T1 1 - T2 Average time to pass the token
- Maximum Throughput
- Frame Time T1
- ------------------------------ ----------
- Time Frame Overhead T1 T2
19Effect of propagation delay
- Effect of propagation delay on throughput
- Case 1 a lt 1 (Packet longer than ring)
- T2 time to pass token to the next station a/N
- Case 2 a gt 1 (Packet shorter than ring)
- Note Sender finishes transmission after T1 1,
but cannot release the token until the token
returns - T1T2 max(1, a) a/N
20Performance of Token Ring
- Given a Token Ring LAN with a 2.5 km long
- Assume average packet size is 620 bits
- Assume, we have 40 computers connected
- Assume it is a 10-Mbps LAN
- Length of Ring in bits 125 bits 40 (delay)
bits 165 bits - Efficiency Ttime / (Ttime propagation delay
/N) - Propagation delay 1.25 x 10-5 sec
- Ttime 6.2 x 10-5 sec
- Efficiency 99.5
21Average Delay of Token Ring
- Given a Token Ring LAN with a 2.5 km long
- Assume average packet size is 620 bits
- Assume, we have 40 computers connected
- Assume it is a 10-Mbps LAN
- Assume on average the token is half way around
the ring. - On average each computer will have to wait for 20
other computers to finish - Average delay 20 (Ttime propagation delay/40)
1.24 10-3 sec - Worse Case delay 40 (Ttime propagation
delay/40) 2.48 10-3 sec
22Performance of Token Ring
- Given a Token Ring LAN with a 2.5 km long
- Assume average packet size is 620 bits
- Assume, we have 40 computers connected
- Assume it is a 100-Mbps LAN
- Length of Ring in bits 1250 bits 40 (delay)
bits 1290 bits - Efficiency Ttime / (propagation delay
propagation delay /N) - Propagation delay 1.25 x 10-5 sec
- Ttime 6.2 x 10-6 sec
- Efficiency 48.4
23dual attach node
c o n c e n t r a t o r
dual attach node
single attach nodes
FDDI Dual Ring Structure
243
4
Bridge to Ethernet LAN
5
node 1
Concentrator
2
6
Concentrator
7
10
Bridge to token- ring LAN
8
9
An example FDDI topology (only primary ring is
shown)
25FDDI MAC Operation
- When a token arrives each station follows this
procedure - THT TTRT TRT
- TRT 0
- Send Synchronous Data
- IF THT gt 0, enable THT and start sending
Asynchronous data as long as THT gt 0
26FDDI MAC Example-1
27FDDI MAC Example-1
28FDDI MAC Example-2
29FDDI MAC Example-2
30FDDI MAC Example-3
31FDDI MAC Example-3
32FDDI MAC Example-Sync. (TTRT 80)
Maximum Throughput 80 / 84 95.23
33FDDI MAC Example-Asyn. (TTRT100)
34FDDI MAC Example(Asyn. TTRT100)
Time 404, station 1 gets the token
35Max Throughput
- Example n 4, TTRT 100, D 4
- Max. Throughput 384 / 404
- Max. Access DelayT(n-1)2D 308 (Station 4) in
the above example