(i) Hubs
An Ethernet hub (or a concentrator) is the central device in a network segment that connects all nodes in the segment. It receives signals on one of its ports and retransmits them to all other ports except the receiving port. It is also known as a multi port repeater. Hubs work at the Physical layer (Layer 1) of the OSI model. Since a hub cannot decide the destination port, it is considered an inefficient device. In a typical implementation, UTP cables are used to connect nodes (computers or printers) to hubs. Hubs can be cascaded (joined together) to extend the network segment. Most ports of the hub use RJ-45 connectors, but AUI and BNC connectors are also provided to extend the segment to legacy 10Base2 and 10Base5 networks. The two types of hubs are described here:
Active hub
An active hub receives signals at its ports and regenerates them before passing them onto all other ports. However, it does not perform any processing in terms of error checking.
Passive hub
A passive hub acts as a simple gateway for incoming signals and does not regenerate them before passing them onto other ports.
Ethernet hubs are available in a variety of sizes and costs, depending on the number of ports. Smaller hubs with 4, 8, or 12 ports are known as workgroup hubs, while hubs with 24 or 32 ports are known as high-density hubs.
An Ethernet hub (or a concentrator) is the central device in a network segment that connects all nodes in the segment. It receives signals on one of its ports and retransmits them to all other ports except the receiving port. It is also known as a multi port repeater. Hubs work at the Physical layer (Layer 1) of the OSI model. Since a hub cannot decide the destination port, it is considered an inefficient device. In a typical implementation, UTP cables are used to connect nodes (computers or printers) to hubs. Hubs can be cascaded (joined together) to extend the network segment. Most ports of the hub use RJ-45 connectors, but AUI and BNC connectors are also provided to extend the segment to legacy 10Base2 and 10Base5 networks. The two types of hubs are described here:
Active hub
An active hub receives signals at its ports and regenerates them before passing them onto all other ports. However, it does not perform any processing in terms of error checking.
Passive hub
A passive hub acts as a simple gateway for incoming signals and does not regenerate them before passing them onto other ports.
Ethernet hubs are available in a variety of sizes and costs, depending on the number of ports. Smaller hubs with 4, 8, or 12 ports are known as workgroup hubs, while hubs with 24 or 32 ports are known as high-density hubs.
(ii) Switches
Like a hub, a switch is also the central device that connects multiple nodes in a Network segment using UTP or STP cables. But unlike the hub that sends the received signal to every port, a switch sends the signal only to the destination node. A switch is an intelligent device that learns the MAC address of the destination from the data packet and sends the packet to the intended node only. This results in data direct communication between two nodes, improved network performance, and reduced collisions.
Switches work at the Data Link layer (Layer 2) of the OSI networking model. Switches can work in a full-a mode, which is a mode that enables nodes to transmit and receive data simultaneously. Thus a 100 Mbps switch working in a full-duplex mode can provide 200 Mbps data transfer speed. Switches are preferred in large networks where hubs can become a bottleneck for network performance. Figure
Like a hub, a switch is also the central device that connects multiple nodes in a Network segment using UTP or STP cables. But unlike the hub that sends the received signal to every port, a switch sends the signal only to the destination node. A switch is an intelligent device that learns the MAC address of the destination from the data packet and sends the packet to the intended node only. This results in data direct communication between two nodes, improved network performance, and reduced collisions.
Switches work at the Data Link layer (Layer 2) of the OSI networking model. Switches can work in a full-a mode, which is a mode that enables nodes to transmit and receive data simultaneously. Thus a 100 Mbps switch working in a full-duplex mode can provide 200 Mbps data transfer speed. Switches are preferred in large networks where hubs can become a bottleneck for network performance. Figure
Switches forward data packets using one of the following forwarding techniques:
Cut Through
The switch reads only the hardware address from the data frame and starts sending it to the destination. It does not perform any error checking. This improves speed. A switch using a cut-through technique may fall back to the store-and-forward technique if it finds that the destination port is busy at the time of transmission.
Store-and-Forward
The switch stores the entire packet in its memory buffer and performs error checking. This prevents forwarding of errors onto the rest of the network. This method is slower than the cut-though method due to the error-checking process, and affects network performance.
Fragment Free
The switch takes advantage of both cut-through and store-and-forward techniques. It reads the first 64 bytes of the frame and leaves error checking to the next device working at the upper layers of OSI model.
Cut Through
The switch reads only the hardware address from the data frame and starts sending it to the destination. It does not perform any error checking. This improves speed. A switch using a cut-through technique may fall back to the store-and-forward technique if it finds that the destination port is busy at the time of transmission.
Store-and-Forward
The switch stores the entire packet in its memory buffer and performs error checking. This prevents forwarding of errors onto the rest of the network. This method is slower than the cut-though method due to the error-checking process, and affects network performance.
Fragment Free
The switch takes advantage of both cut-through and store-and-forward techniques. It reads the first 64 bytes of the frame and leaves error checking to the next device working at the upper layers of OSI model.
(iii) Media Access Unit (MAU)
An MAU—also called Multi-Station Access Unit (MSAU)—is used in Token Ring networks as a central device that connects all nodes in the network segment. This is equivalent to using a hub or a switch in Ethernet networks and results in giving the network a physical star look, though its logical topology remains a ring. Multiple MAUs can be connected using the Ring In (RI) and Ring Out (RO) ports in order to extend the network. The RO port of one MAU is connected to the RI port of the second MAU, and so on. The RO port of the last MAU is connected back to the RI port of the first MAU in the network to complete the ring.
An MAU—also called Multi-Station Access Unit (MSAU)—is used in Token Ring networks as a central device that connects all nodes in the network segment. This is equivalent to using a hub or a switch in Ethernet networks and results in giving the network a physical star look, though its logical topology remains a ring. Multiple MAUs can be connected using the Ring In (RI) and Ring Out (RO) ports in order to extend the network. The RO port of one MAU is connected to the RI port of the second MAU, and so on. The RO port of the last MAU is connected back to the RI port of the first MAU in the network to complete the ring.
(iv) Bridges
A network bridge is used for two purposes: connecting two LAN segments to form a larger segment and dividing a large network segment into smaller segments. It works at the Data Link layer (Layer 2) of the OSI model. Like network switches, bridges also learn the MAC address of devices and forward data packets based on the destination MAC address. In older bridges, the MAC addresses had to be defined manually, and it tooka significant amount of time to configure a bridge. Most of the newer bridges can dynamically build lists of MAC addresses by analyzing data frames. These bridges are called learning bridges, due to this advanced functionality. Most of the functionality of bridges is now available in switches. Hence, they are rarely used in networks these days. Bridges fall into the following categories:
A network bridge is used for two purposes: connecting two LAN segments to form a larger segment and dividing a large network segment into smaller segments. It works at the Data Link layer (Layer 2) of the OSI model. Like network switches, bridges also learn the MAC address of devices and forward data packets based on the destination MAC address. In older bridges, the MAC addresses had to be defined manually, and it tooka significant amount of time to configure a bridge. Most of the newer bridges can dynamically build lists of MAC addresses by analyzing data frames. These bridges are called learning bridges, due to this advanced functionality. Most of the functionality of bridges is now available in switches. Hence, they are rarely used in networks these days. Bridges fall into the following categories:
Transparent Bridge
This bridge forwards data packets to the destination network segment by reading the destination MAC address. The network devices are unaware of the presence of the bridge. This bridge builds the MAC address table as it receives data packets. If the bridge does not find a destination MAC address in its list, it floods all ports with the data packet except the source port.
Source Route Bridge
This bridge is used in Token Ring networks. The bridge uses two frame types to find the route for the data: a Source Route (SR) frame and an All-Route (AR) frame.
Translation Bridge
This bridge is used to connect two network segments that use different protocols at the Data Link layer. For example, a translation bridge can join a Token Ring network to an Ethernet network or an FDDI to a Token Ring.
Spanning tree protocol.
This bridge forwards data packets to the destination network segment by reading the destination MAC address. The network devices are unaware of the presence of the bridge. This bridge builds the MAC address table as it receives data packets. If the bridge does not find a destination MAC address in its list, it floods all ports with the data packet except the source port.
Source Route Bridge
This bridge is used in Token Ring networks. The bridge uses two frame types to find the route for the data: a Source Route (SR) frame and an All-Route (AR) frame.
Translation Bridge
This bridge is used to connect two network segments that use different protocols at the Data Link layer. For example, a translation bridge can join a Token Ring network to an Ethernet network or an FDDI to a Token Ring.
Spanning tree protocol.
The problem with bridges is that they cannot be used for large networks. When multiple bridges are used in a large network, they start confusing each other. This results in bridging loops, a term used when one bridge makes the other bridge believe that a device is located in a network segment—while it actually is not. To overcome the bridging loops problem, bridges use the spanning tree protocol. This is defined in the IEEE 802.1d standard. Using this protocol, the bridge interfaces are assigned a value that helps control the way bridges learn MAC addresses and disable inactive or non existing links.
(v) Routers
Routers are used to connect two or more network segments. These devices work on the Network layer (Layer 3) of the OSI model. Routers use Internet Protocol (IP) addresses to determine the source and destination of the data packet. Typically, routers receive the data packet, determine the destination IP address, and forward the packet to the next hop, which may either be the final destination of the packet or another router on the path. Routers can be implemented as a software service or as a dedicated hardware device. A wired or wireless router in a home network is an example of a small network router that connects the home network to the ISP’s network. Microsoft’s Routing and Remote Access Service (RRAS) is an example of a software router. A Windows Server 2000/2003 computer with at least two network interface cards can be configured as a router to connect network segments. Figure
Routers are used to connect two or more network segments. These devices work on the Network layer (Layer 3) of the OSI model. Routers use Internet Protocol (IP) addresses to determine the source and destination of the data packet. Typically, routers receive the data packet, determine the destination IP address, and forward the packet to the next hop, which may either be the final destination of the packet or another router on the path. Routers can be implemented as a software service or as a dedicated hardware device. A wired or wireless router in a home network is an example of a small network router that connects the home network to the ISP’s network. Microsoft’s Routing and Remote Access Service (RRAS) is an example of a software router. A Windows Server 2000/2003 computer with at least two network interface cards can be configured as a router to connect network segments. Figure
Routers communicate to each other using routing protocols. They maintain a list of IP addresses in routing tables. Routing tables can be built statically or dynamically as discussed in the following list:
Static routing
When static routing is used, administrators manually configure routing tables by entering appropriate routing information. This method works only for very small networks. In large networks, it is very difficult to manually configure routing tables. As the routing tables grow or there is a change in the network, the routing tables must be updated manually. The process is time consuming and error-prone.
Dynamic routing
Routers use dynamic routing protocols when working in a dynamic routing environment. Dynamic routing protocols enable routers to get routing information from other routers and advertise their own routing information in order to build and maintain routing tables. Dynamic routing protocols fall into two categories, discussed next.
Static routing
When static routing is used, administrators manually configure routing tables by entering appropriate routing information. This method works only for very small networks. In large networks, it is very difficult to manually configure routing tables. As the routing tables grow or there is a change in the network, the routing tables must be updated manually. The process is time consuming and error-prone.
Dynamic routing
Routers use dynamic routing protocols when working in a dynamic routing environment. Dynamic routing protocols enable routers to get routing information from other routers and advertise their own routing information in order to build and maintain routing tables. Dynamic routing protocols fall into two categories, discussed next.
Distance vector routing protocol.
A distance vector routing protocol assumes that the network is made up of several routers. Routers using this protocol depend on other routers to advertise their routing information periodically. These advertisements (or updates) are typically sent every 30 seconds. Routers can also be configured to send triggered updates when they detect any change in network topology.
RIPv1 (Routing Information Protocol version 1) and RIPv2 (Routing Information Protocol version 2) are distance vector protocols that work on the principle of hop count. RIPv1 works only on TCP/IP networks, while RIPv2 works on both TCP/IP and IPX/SPX networks. The RIP version that supports IPX is sometimes called IPX RIP also. A hop is a value assigned to each router on the way to the final destination. RIP supports a maximum of 15 hops in the network. A destination beyond 15 hops is considered unreachable. The following are the main disadvantages of distance vector routing protocols:
• Periodic update is a slow process that affects network performance.
•Periodic updates generate considerable network traffic, making the protocol inefficient on large networks.
• Routing loops are created when routers advertise incorrect routing information.
RIPv1 (Routing Information Protocol version 1) and RIPv2 (Routing Information Protocol version 2) are distance vector protocols that work on the principle of hop count. RIPv1 works only on TCP/IP networks, while RIPv2 works on both TCP/IP and IPX/SPX networks. The RIP version that supports IPX is sometimes called IPX RIP also. A hop is a value assigned to each router on the way to the final destination. RIP supports a maximum of 15 hops in the network. A destination beyond 15 hops is considered unreachable. The following are the main disadvantages of distance vector routing protocols:
• Periodic update is a slow process that affects network performance.
•Periodic updates generate considerable network traffic, making the protocol inefficient on large networks.
• Routing loops are created when routers advertise incorrect routing information.
There are two methods to get around the routing loops problem in distance vector protocols. The first method is split horizon, which prevents a router from advertising a route to the same router from which it received the route information. The second method is poison reverse, which advertises back the route it learns from a router with a hop count of 16 (unreachable).
Link state routing protocol.
Link state routing protocol.
Link state routing protocols use Link State Advertisements (LSA) to update routing tables. The LSA is a data packet that contains routing information about the sending router only. This packet is sent to all routers in the network so that other routers can build routing tables. This is in contrast with the distance routing protocols where all routers advertise their entire routing tables to all other routers, thus generating significant amount of network traffic.
Open Shortest Path First (OSPF) and NetWare Link State Protocol (NLSP) are examples of link state routing protocols. The link state routing protocols are best suited for large networks, as there is no limit such as the hop count. They keep update traffic to the minimum and can correct the routing table information quickly if there is a change in the network topology. This characteristic is known as convergence.
Open Shortest Path First (OSPF) and NetWare Link State Protocol (NLSP) are examples of link state routing protocols. The link state routing protocols are best suited for large networks, as there is no limit such as the hop count. They keep update traffic to the minimum and can correct the routing table information quickly if there is a change in the network topology. This characteristic is known as convergence.
(vi) Gateways
In computer networks, a gateway is a device that translates one format of data packets to another format. They are also called protocol translators. A router connecting two different types of network segments or a bridge connecting two network segments using different Layer 2 protocols are examples of gateways. Gateways are necessary to provide interoperability between two distinct network formats. It is notable that gateways only convert (translate) data formats, but the that data itself remains unchanged.
(vii) Channel Service Unit (CSU)/Data Service Unit (DSU)
A CSU/DSU is a digital interface device that connects a local area network to a wide area network. Typically, the CSU/DSU is installed between a LAN and the access point provided by the provider of the WAN service. Most of the newer routers now include the functionality of CSU/DSU. For example, a router connecting a LAN to the Internet is also functioning as a CSU/DSU unit.
In computer networks, a gateway is a device that translates one format of data packets to another format. They are also called protocol translators. A router connecting two different types of network segments or a bridge connecting two network segments using different Layer 2 protocols are examples of gateways. Gateways are necessary to provide interoperability between two distinct network formats. It is notable that gateways only convert (translate) data formats, but the that data itself remains unchanged.
(vii) Channel Service Unit (CSU)/Data Service Unit (DSU)
A CSU/DSU is a digital interface device that connects a local area network to a wide area network. Typically, the CSU/DSU is installed between a LAN and the access point provided by the provider of the WAN service. Most of the newer routers now include the functionality of CSU/DSU. For example, a router connecting a LAN to the Internet is also functioning as a CSU/DSU unit.
No comments:
Post a Comment