Figure 6.1 An Ethernet Network interface Card
Introduction
Network
Interface Card
Physical Layer
Data Link
Hubs
Older Implementations
of Ethernet
Repeaters
Updating Ethernet using
Multiport Repeaters
Switched Hubs
Layer 2 switching
Switch Connections
Mathematical
Analysis of a Switched Hub
Conclusion
Resources
Appendix
Description of Terms
Used
Netgear Switch
Specifications
3Com Multiport Repeater
Specifications
This lecture will examine:
Network Interface Cards
Old style Ethernet
Repeaters
Hubs and their various implementations
Layer 2 switching
Network Interface Card NIC (LAN Adapter)
This is a circuitboard (measuring around 10cm by 7 cm)
that is inserted into a free PCI or ISA
slot within a PC located on the PCs motherboard. The NIC may be very small,
around the size of a credit card, on a Laptop PC.
Figure 6.1 An Ethernet Network
interface Card
A NIC is an expansion card that contains all the circuitry
necessary to connect a PC to a network. In order to do this, the NIC must
be able to carry out all the processes that are defined in the OSI
model. The NICS are classified as:
The card must provide an interface for plugging cables into itself. This may be coaxial cable on older cards or twisted pair on newer cards using RJ45 connectors. It must also be able to provide the signals necessary for transmission onto the transmission medium. Different physical layer implementations may allow for optical, radio or infra red connections to the card.
The card contains the MAC (hardware) address of the card which must be unique within a particular local area network. This is to identify the piece of equipment uniquely within the network. It also implements the MAC method used e.g. CSMA/ CD.
The NIC must also implement packet formation, parallel to serial conversion, serial to parallel conversion, encoding and decoding, handshaking, buffering etc.
The actual functions and layers in use will depend on the protocol stack in use.
The term HUB is used to denote a piece of equipment which is found at the centre of a series of cables and can provide connections between the computers attached to the opposite ends of these cables. There are many pieces of networking equipment that may be referred to as 'hubs' but this is a loose generic term as the operation of the different types of hubs varies greatly.
Hubs may be used to connect several networked components e.g. servers, PCs, workstations together.
General characteristics of a hub are that they have a
number of RJ-45 ports or sockets into which networked components may be
plugged by means of a cable. The hubs we have in the lab have RJ-45 sockets
to plug in 10 or 100BaseT
connections.
Figure 6-2 A Netgear
4-port Hub.
It is insufficient to define the equipment shown above
as a hub. To be properly precise, it is a Multiport Repeater. The manufacturer's
page is online.
Older Implementations of Ethernet
In early Ethernet (IEE802.3) networks, the characteristic layout was a common coaxial cable running between all the networked components. This was known as a Bus. The Bus ran throughout the building and all PCs and servers were connected to it. Any PC that wished to send a message placed the message onto the Bus which carried the message along the entire length of the cable. All other equipment connected to the Bus was able to read the message. This caused security concerns in some instalations as all messages reached all other machines.
Figure 6-3 Several PCs and
a Server share a Bus Network
The backbone of the network was the Bus, being the shared medium through which all signals passed. If the bus was damaged, the network would malfunction.
Today, the only time that you will see a Bus implementation
of an Ethernet system is in legacy (old) installations and out of date
text books.
Digital repeaters operate at the Physical layer (layer 1) of the OSI model. They do not examine the contents of the data upon arrival nor perform any error-checking over the data.
The function of a digital repeater is to regenerate and amplify all incoming signals and pass them to the output port. The newly re-transmitted signal will not contain noise or unwanted aspects of the signal which may have been introduced during the signal’s journey to the repeater. Repeaters are necessary to overcome the loss of strength (attenuation) as signals travel along cables or propagate through space.
Figure 6-4 A two port digital
repeater is able to 'clean up' a signal and retransmit again with full
power
Note: the received signal may be on an electromagnetic or optical transmission medium using wire, fibre or wireless media.
Bits that arrive at the repeater are firstly decoded then
re-encoded and transmitted onto all outbound ports after a short delay,
approximately 1 bit time. Published times for retransmission
are in the order of 10 to 40 microseconds. To function, there is no need
for the repeater to understand the contents of the frames that it is forwarding.
Updating Ethernet using Multiport Repeaters
A Multiport Repeater is a repeater that has several sockets for plugging in data cabling. Any data arriving at any one of the ports is re-transmitted on all other ports at full strength as described above.
Figure 6.5 A Netgear
4-port repeater
A multiport repeater (hub) is able to replace the bus as a component into which all the network components may be connected. Its electrical characteristics have been designed so that it appears to the servers and PCs exactly as a bus would i.e. its impedance (how it appears from an electrical point of view) is the same as that of the bus. However, instead of having a bus running throughout a room or building, the networked equipment is plugged directly into the hub. It is as though the cabling that formed the bus has been collapsed into the hub.
To
another hub or networked component
Figure 6.6 PCs and Server connected by Multiport Repeater (Hub)
The multiport repeater may now be placed in a secure area or room or locked in a cabinet to prevent any unauthorised access. This improves the security aspect of the installation leaving just the cabling from the PCs to the hub exposed as a security risk.
Hubs often have LED indicators on their front panel to indicate the the status of the separate ports on the hub. CSMA/CD is still used here because the hub just broadcasts any received messages on all ports to all connected equipment. Therefore there is still the same risk of data collision as there was with a common Bus.
Multiport repeaters may have different data rate ports and these will generally be printed on the front of the hub adjacent to the ports. A 10/100 Hub is able to operate at either 10 or 100 Mbits/ second. The hub is able to sense from the NIC that is transmitting the appropriate speed to use. When two nodes having different data rates are communicating via the hub, the hub's port data rate will be determined by the slower of the two machines' NICs.
Figure 6.7 RJ-45 ports on
a Hewlett Packard switched hub
Switched hubs are more commonly referred to as switches. This is misleading as there are two completely different types of switches available, Layer 2 and Layer 3.
The switches that are discussed in this section of the notes are Layer 2 switches. These are of a similar type to those that are in use in A150, where the NIC laboratory took place.
The operation of a layer 2 switch depends upon MAC (hardware) addresses that are held on NICs. As devices are connected to switch ports, the switch interrogates the NIC on the far end of the cable and determines the MAC address of the attached device(s).
After all connected devices have made one transmission via the switch, memory within the switch will have stored all MAC addresses that are attached and also mapped them to the port to which they are attached.
When a node wishes to send a frame to another node, the MAC address of the receiving node is examined by the switch and providing that the intended receivee is not busy, the switch will make a connection between the pair of ports to which the two devices involved in the communication are attached.
See specifications of a simple Netgear Layer 2 switch.
A switched hub performs a similar function to that of a multiport repeater, i.e. it may be used as a connector for network equipment that may be interconnected using Category 5 cabling with RJ-45 connectors. It also cleans up and boosts the power of received signals.
The big difference is that a switch is able to connect just the two endpoints that wish to communicate. In the diagram below, the pairs of PCs, A, B and C may communicate simultaneously because they are connected directly to each other on demand by the switch. The different colours used for the cables denote communications taking place between pairs of nodes.
Compare this with a multiport repeater which cannot deal
with more than one message at a time.
Figure 6.8 PCs and Server
connected using a Switch
The diagram above helps to show that the potential information carrying capacity of the network has increased because several pairs of devices can communicate simultaneously.
Mathematical Analysis of a Switched Hub
Given that a switch may connect pairs of ports together independently of other messages being transmitted by different equipment on the segment, we may calculate e the extra bandwidth that may be gained by deploying a switched hub in place of a multiport repeater.
If the PCs were fitted with 10 Mbps NICs, the maximum data transfer rate using a Bus or non-switched hub would be (theoretically) 10 Mbps (this figure is never reached in practice).
However, in the example above with a switched hub, three pairs of PCs are communicating at the same time and there are still two ports unused. Thus the new network carrying capacity is:
Number of ports X
NIC speed (Mbits/ second)
2
Here this works out as 4 X 10 Mbps = 40 Mbps.
Thus the 8 port switch has increased the potential information carrying capacity of the network by a factor of 4.
The switches in use today are often remotely manageable
units. This means that they may be configured from a remote location. It
is possible to enable or disable ports, view their usage and statistics
etc. without having to be on the premises of the switch.
Figures 6.9 and 6.10 depict
the external display panel of a switched hub which can show port usage.
The direct connections between the end-points does mean that CSMA/ CD is not strictly needed here, but it has been retained however because other parts of the network may not be switched.
Privacy is assured with a switch because the only parties that can "hear" the message are the two end points that are involved in the data exchange. This may be a very useful and desirable property of a switch in some networking situations.
The NIC contains the hardware (MAC) address of the card. It has connections to join the physical layer (cabling, media etc.) to the card for transmissions to take place.
The NIC also deals with framing and MAC addressing.
A hub is a loose generic term that can describe network equipment into which several other communication devices may be plugged.
The function of a repeater is to remove unwanted noise from an incoming signal, boost the signal's strength to full power and retransmit the signal once more.
One implementation of a hub is a multiport repeater. This carries out the function described above but may have many ports to plug network devices into. When one device transmits, the multiport repeater receives the signal and re-transmits the signal on all of the other ports. This replaces the cabling that ran throughout the installation known as the bus.
A second implementation is a switched hub. This is more
commonly known as a switch. In LAN segments, Layer 2 switches are used.
These devices form individual connections between pairs of its ports, the
connections being made on the basis of the MAC addresses of the device(s)
attached to the ports.
10BaseT
uses unshielded twisted pair. May use Category 3 or Category 5 cabling
100BaseT
uses unshielded twisted pair. May use Category 5 or Category 5A cabling
CSMA/ CD
Carrier Sense with Multiple Access; Collision Detection
OSI
Open Systems Interconnect, a communication model developed to help promote
interconnection of different manufacturer's network equipment.
PCI
Peripheral Component Interface slot, these are white slots on the
motherboard and are capable of higher data transfer speeds than ISA
ISA
Industry Standard Architecture, these are black slots in the motherboard
and are slower than PCI
Enhances productivity and network reliability by segmenting the workgroup into smaller units.
Eight auto-sensing UTP ports
One speed-sensing UTP port selectable between Normal or Uplink connection.
Stores up to 4,096 MAC addresses per system.
Half/full duplex switch
Filtering and Forwarding Rates
14,800 packets per second for 10 Mbps ports
148,000 packets per second for 100 Mbps ports
Specifications of a 3Com Multiport Repeater (from www.3com.com)
The Dual Speed Hub 500 is an autosensing 10/100 Mbps hub that complements PS Hub 10 Mbps Ethernet hubs. When you need a stackable solution that delivers 10/100 Fast Ethernet connectivity to your network, mix and match the PS Hub family with Dual Speed Hub 500s for a winning combination. The 10/100 Mbps autosensing capability on each port supports 10 Mbps, automatically detecting the speed of the attached PC network interface card (NIC). Designed for the smoothest and most cost-effective migration to Fast Ethernet, the SuperStack II Dual Speed Hub 500 works with existing 10 Mbps NICs, and when used with 10/100 Mbps NICs, automatically upgrades the connection to 100 Mbps. Plus users can add SNMP/RMON management, when required. The Dual Speed Hub 500 stacks with SuperStack II PS Hub 40/50s, SuperStack II Hub 10s, and LinkBuilder® FMS II hubs, offering the most flexible solution available on the market.
Features 12 or 24 RJ-45 ports supporting 10BASE-T transmission over two pairs of Category 3, 4, or 5 UTP cable, or 100BASE-TX transmission over two pairs of Category 5 UTP/STP cable
Dual Speed Hub 500
4 K maximum addresses
Learning rate: 50 addresses/second maximum
Aging period default: 5 days
Filtering: 163,680 frames/second
Forwarding:
10 Mbps—›100 Mbps: 14,880 frames/second
100 Mbps—›10 Mbps: 14,800 frames/second
Latency (store and forward only):‹—<=40 µS
PS Hub 50
512 maximum addresses
Forwarding:
10 Mbps—›10 Mbps: 14,800 frames/second
10 Mbps—›100 Mbps Downlink: 14,800 frames/second
100 Mbps Downlink—›10 Mbps: 14,800 frames/second
Latency (store and forward only): 8 µS
Back to top of page
© MM Clements 2001