1
00:00:00,240 --> 00:00:02,430
Now let's look at a mac address in more detail.

2
00:00:02,460 --> 00:00:05,000
It's once again six bytes in length.

3
00:00:05,040 --> 00:00:09,060
And if you remember, a byte is eight bits in length.

4
00:00:09,060 --> 00:00:12,390
So six times eight gives you 48 bits.

5
00:00:13,140 --> 00:00:18,150
Three bytes or 24 bits is the UI portion of the address.

6
00:00:18,540 --> 00:00:20,760
Three bytes or 24 bits.

7
00:00:20,790 --> 00:00:26,880
Is network interface card specific and is the unique identifier of that network interface card.

8
00:00:27,560 --> 00:00:33,380
Now in the UI portion in the first octet or most significant octet.

9
00:00:33,410 --> 00:00:38,360
In other words, the first byte in the UI, the least significant bit.

10
00:00:38,390 --> 00:00:45,560
In other words, the last bit of the first octet or first byte is either set to zero, which indicates

11
00:00:45,560 --> 00:00:51,110
unicast or it's set to one which indicates multicast unicast.

12
00:00:51,110 --> 00:00:57,470
Traffic, if you remember, is a conversation between two devices where one device is sending the traffic

13
00:00:57,470 --> 00:00:59,930
and the other device is receiving the traffic.

14
00:00:59,960 --> 00:01:02,910
So device A is talking to device B.

15
00:01:02,930 --> 00:01:11,090
Multicast is where one device is sending traffic to multiple devices that have subscribed to the multicast.

16
00:01:11,360 --> 00:01:16,520
Now, this makes it very efficient for Ethernet switches to know whether they should flood the frame

17
00:01:16,520 --> 00:01:17,630
out of all ports.

18
00:01:17,630 --> 00:01:24,080
When multicast traffic is received by a layer two switch, that traffic is flooded out of all ports,

19
00:01:24,080 --> 00:01:27,380
whereas unicast traffic is typically not flooded.

20
00:01:27,740 --> 00:01:34,760
So by reading the bit in the frame, the layer two switch knows how to process traffic.

21
00:01:34,970 --> 00:01:38,210
The second least significant bit in the first octet.

22
00:01:38,210 --> 00:01:44,390
So in other words, we still looking at the first octet, but second least significant bit is either

23
00:01:44,390 --> 00:01:51,980
set to zero, which means that it's a globally unique Mac address or it's set to one, which means that

24
00:01:51,980 --> 00:01:54,770
an administrator has changed the MAC address.

25
00:01:54,770 --> 00:02:01,130
So that would be for the example that I did previously where I changed the Mac address on my PC.

26
00:02:01,160 --> 00:02:10,970
The Zero means it's a unique Mac address designated by a manufacturer, whereas a one means that a administrator

27
00:02:10,970 --> 00:02:14,630
locally changed the Mac address of the interface.

28
00:02:14,840 --> 00:02:22,700
Now in Ethernet, when a bus topology is used, devices use what's called carrier sense, multiple access

29
00:02:22,700 --> 00:02:26,930
slash collision detection or CSA slash CD.

30
00:02:27,260 --> 00:02:34,040
This operates as follows When a device wants to send traffic, it should first check to hear if any

31
00:02:34,040 --> 00:02:39,500
other device is speaking so the device will not communicate onto the network.

32
00:02:39,500 --> 00:02:44,000
If it hears another device that's called Carrier Sense.

33
00:02:44,030 --> 00:02:50,270
Carrier Sense is essentially sensing the network to hear if another device is speaking.

34
00:02:50,270 --> 00:02:57,710
Multiple access means that any device can communicate across that segment as long as no other device

35
00:02:57,710 --> 00:02:58,820
is communicating.

36
00:02:59,240 --> 00:03:07,640
Now, this is different to the old mainframe days where a central device would pull terminals to allow

37
00:03:07,640 --> 00:03:08,840
them to communicate.

38
00:03:08,990 --> 00:03:15,890
In Ethernet, we're using a distributed environment where each device can independently communicate

39
00:03:15,890 --> 00:03:19,610
across the network without permission from other devices.

40
00:03:19,610 --> 00:03:26,720
However, a device should only send traffic if no other device is speaking, and that's because we want

41
00:03:26,720 --> 00:03:29,630
to avoid collisions in an Ethernet environment.

42
00:03:29,810 --> 00:03:37,040
As another analogy, when traditional telephones are connected to a PBX, the PBX is in charge of the

43
00:03:37,040 --> 00:03:38,180
communications.

44
00:03:38,360 --> 00:03:41,210
That's not true in an Ethernet environment.

45
00:03:41,210 --> 00:03:44,840
Every device is independent of other devices.

46
00:03:44,930 --> 00:03:51,050
However, if collisions do take place, there's an option in Ethernet to detect collisions.

47
00:03:51,050 --> 00:03:58,580
When a device detects that a collision has taken place, it may send a back off or jamming signal to

48
00:03:58,580 --> 00:04:02,180
indicate that a collision has taken place once again.

49
00:04:02,180 --> 00:04:08,660
In this environment, terminates are used at the end of the cable to ensure that signals don't bounce

50
00:04:08,660 --> 00:04:11,450
back, causing additional collisions.

51
00:04:11,630 --> 00:04:19,880
Now, in a given scenario, it may happen that two devices want to communicate at exactly the same time,

52
00:04:19,880 --> 00:04:23,600
but at that point in time, no devices are speaking.

53
00:04:23,870 --> 00:04:31,880
So let's say that in this example A wants to communicate with C, so A wants to send traffic onto the

54
00:04:31,880 --> 00:04:35,960
network with a source address of a and destination address of C.

55
00:04:35,960 --> 00:04:40,220
But at that exact point in time, D also wants to communicate.

56
00:04:40,400 --> 00:04:46,310
In this case, D wants to communicate with B, so it wants to send a frame onto the network with a source

57
00:04:46,310 --> 00:04:54,920
address of D and destination address of B now in line with C, SMA Slash CD, both A and D.

58
00:04:54,920 --> 00:05:02,780
Firstly, check to see if anyone is speaking so they use carrier sense or S to check the wire.

59
00:05:02,930 --> 00:05:06,800
At this point in time, no device is communicating on the network.

60
00:05:06,800 --> 00:05:14,240
However, because of multiple access, any device can access the cable without permission from any other

61
00:05:14,240 --> 00:05:14,930
device.

62
00:05:14,930 --> 00:05:18,560
So both A and D send traffic onto the network.

63
00:05:18,560 --> 00:05:25,190
But because this is ten base two or in other words, base band, only one signal is allowed across the

64
00:05:25,190 --> 00:05:26,540
wire at any given.

65
00:05:26,630 --> 00:05:27,200
Time.

66
00:05:27,200 --> 00:05:30,440
So in this example, a collision takes place.

67
00:05:30,770 --> 00:05:37,910
Now, if a transmitting data station or PC detects another signal on the wire while transmitting its

68
00:05:37,910 --> 00:05:45,440
frame, it will stop transmitting that frame and then send a jamming signal as well as waiting a random

69
00:05:45,440 --> 00:05:51,620
period of time known as a back off delay before trying to send the signal again.

70
00:05:51,650 --> 00:05:58,010
This will prevent machines or PCs from repeatedly attempting to transmit at the same time.

71
00:05:58,220 --> 00:06:06,560
However, the probability of collisions becomes greater as the cable length increases and as more devices

72
00:06:06,560 --> 00:06:08,150
are added onto the network.

73
00:06:08,180 --> 00:06:15,500
In other words, it's more likely that collisions will take place with longer cable lengths and more

74
00:06:15,500 --> 00:06:16,420
devices.

75
00:06:16,430 --> 00:06:23,570
So as you add more and more devices to this network and extend the cable length, the probability of

76
00:06:23,570 --> 00:06:25,520
collisions increases dramatically.

77
00:06:27,100 --> 00:06:29,590
Now there were other issues with ten BS too.

78
00:06:29,620 --> 00:06:31,960
The first issue is cable length.

79
00:06:32,470 --> 00:06:36,350
The longer the cable, the greater the signal degradation was.

80
00:06:36,370 --> 00:06:42,880
In other words, as your cable increased in length, the more likely it was that one host's signal would

81
00:06:42,880 --> 00:06:44,980
not be received by another host.

82
00:06:45,220 --> 00:06:51,130
The host on one side of the cable might send a signal, but because of degradation, a host at the other

83
00:06:51,130 --> 00:06:55,450
end of the cable may not be able to receive or interpret the signal.

84
00:06:55,480 --> 00:06:58,060
Another problem is cable breaks.

85
00:06:58,450 --> 00:07:02,880
A cable break at any point would cause the entire network to fail.

86
00:07:02,890 --> 00:07:07,720
So if someone accidentally broke the cable at this point, the whole network would fail.

87
00:07:07,750 --> 00:07:11,620
Host A cannot communicate with other devices in the network.

88
00:07:11,620 --> 00:07:14,590
Those devices cannot communicate with host A.

89
00:07:14,620 --> 00:07:20,290
However, because of this cable break, there is no Terminator on the cable.

90
00:07:20,320 --> 00:07:22,120
The cable is also damaged.

91
00:07:22,120 --> 00:07:25,270
So what happens is signals get reflected.

92
00:07:25,300 --> 00:07:32,740
D might send a signal to C, but it's going to continue across the cable and then it's going to be reflected

93
00:07:32,740 --> 00:07:35,250
back causing collisions in the network.

94
00:07:35,260 --> 00:07:41,950
So this network wasn't very robust in that cable breaks could bring down the entire network.

95
00:07:41,980 --> 00:07:43,500
Now it gets worse.

96
00:07:43,510 --> 00:07:48,280
Ten base two implies ten megabits per second Ethernet.

97
00:07:48,400 --> 00:07:53,220
However, this is not ten megabits per second for each device.

98
00:07:53,230 --> 00:07:58,600
It's ten megabits per second shared between all devices on that segment.

99
00:07:58,690 --> 00:08:06,960
In addition, because of collisions, you can only use between 30 and 40%, so you only getting 30 to

100
00:08:06,960 --> 00:08:08,410
40% utilization.

101
00:08:08,590 --> 00:08:12,250
Collisions increase dramatically above that utilization.

102
00:08:12,250 --> 00:08:15,980
So a conservative figure would be 30% utilization.

103
00:08:16,000 --> 00:08:22,090
That means ten megabits per second would be shared between all devices on that segment.

104
00:08:22,090 --> 00:08:24,280
So in this case, we have four devices.

105
00:08:24,280 --> 00:08:32,559
So that means that ten megabits per second divided by four devices, times 30%, only gives you 0.75

106
00:08:32,559 --> 00:08:37,090
megabits per second and not ten megabits per second per device.

107
00:08:37,419 --> 00:08:45,700
This is not ideal because the bandwidth available to your PC is very low, especially in a large network.

108
00:08:45,700 --> 00:08:51,460
So as more devices are added to the network, the bandwidth available to each device is decreased.

109
00:08:51,490 --> 00:08:55,370
This is also known as a single collision domain.

110
00:08:55,390 --> 00:09:01,360
In other words, if a collision takes place at any point in the network, all devices in this network

111
00:09:01,360 --> 00:09:04,600
are affected by that collision and would need to back off.