TCP/ IP model and OSI layers are the two network models used by the internet. In the OSI model, an Ethernet switch which tends to operates at the given Data-Link layer to create a different collision domain for every single switch port. A multilayer switch though may work at more than one OSI layers, including physical, data link, network and even the transport layer. A router on the other hand is considered a layer 3- the Network layer of the TCP/IP model- device since it does most of its routing decisions according to information in the IP packet of layer 3.
The switch, which is basically a modern and more efficient version of the outdated hub, is a network access device. The router is an internetworking device. They are both intermediary devices. The basic functionalities of the processes that run in them are to regenerate and re-transmit data, maintain information about the pathways that exist across the network and internet, inform other devices in case of errors and transmission failures, guide data along substitute pathways when there is a failure in the link, classify and direct messages according to Quality of Service (QoS) priorities and permit or deny data flow based on security settings.
A Local Area Network (LAN) is a network of computers which typically covers one geographical area, delivering services and applications to those within a mutual organizational structure. A switch is used to connect these computers. Switches lack the capability of connecting multiple networks or distributing an internet connection. A Wide Area Network (WAN) is a network of computers covering a broad area and is used in interconnecting two or multiple Local Area Networks. It is usually built using leased lines. A router connects two or more logical subnets and thus is usually connected to two or more LANs or WANs or also a LAN and its Internet Service Provider (ISP). Routers are also used to link segments of a LAN that has been sub-netted into smaller segments. Routers are located at gateways, the meeting point of two or more networks (Ilascu, 2007)
Packet-switched computer networks employ a forwarding technique called bridging. Bridging is dependent on flooding and source address examination in received packet headers to establish the address location of unknown devices. Switches maintain a table of Ethernet MAC addresses referred to as a CAM Table also known as Bridge forwarding table using a method called Transparent Bridging involving learning, flooding, filtering, forwarding and aging. The switch records the source addresses in entries in the table, while destination addresses are looked up in the table and matched to the proper segment to send the frame. The show mac-address-table command can be used to show this table on a Cisco switch. Source route bridging is another form of bridging that was developed for token ring networks. Routers direct packets according to information stored in the IP routing table. A router looks inside each TCP/IP packet it has received to identify the IP addresses of the source and target, searches its own routing table for an address that best matches between the destination IP address of the network and those in the routing table packet, then forwards the packets as desired to ensure that data reaches its destination. The show ip route command can be used to show the table on a Cisco router. (Cisco, 2009)
Switches and routers are often ignored as security devices because they were originally made to enhance network connectivity rather than network security. Consequently, they are conceptually less secure. An attack on the edge router can cut off the entire network from its users. Weak passwords, unneeded services, protocol and IOS vulnerabilities and IOS configuration errors can cause security breaches. Most routers nowadays have a hardware firewall integrated in their systems thus this has made routers an important component in fostering network security.
There are several technical terms about switches and routers that one would encounter:
i. Ports and interfaces
A port is a physical access point or physical interface connecting a circuit and a device or system where signals are inserted or extracted. It can also be a logical connection that is identified by a protocol address in the packet header, associated with TCP or UDP service. A socket is identified by the IP address and the port number. An interface physically and logically interconnects two other devices or systems
ii. Broadcast, unicast and multicast
Multicast is a transmission mode in which a signal or packet is sent to multiple devices or computers, but not all on a network, unicast is where a single packet or signal is sent to a single device and broadcast is where it is sent to all stations on a network in order to update the routing table.
This is where one network meets another, for example where a LAN and a WAN are joined by a router.
These are rules and conventions used to exchange information between computers or across computer internetworks. Their two major functions are handshaking and line discipline.
iv. Encapsulation and Decapsulation
Different types of information are added as data from the Application layer passes down the protocol stack. A new header in each level and a new name is given to the data enclosed. Decapsulation is the reverse. As data goes up the protocol stack, the Data link layer takes it from the physical medium does a Cyclic Redundancy Check, strips off the respective headers and the rest of the data is passed to an upper layer till it reaches the application layer.
LAN switches depend on packet-switching. A connection between two segments is established long enough to direct out the current packet. Inbound packets are stored in a temporary memory area, buffer; MAC address in the header is read and compared to those in the switch's lookup table. An Ethernet frame comprises a normal packet as the payload of the frame, which has a special header including the MAC address information.
Traffic-routing in a switch
Three methods are used by packet-based switches for traffic-routing. As soon as the packet is detected by the switch, Cut-through switches read the MAC address. The 6 bytes that contain the address information are stored and instantly forwarding the packet to the destination node starts as the rest of the packet comes into the switch. In store-and-forward, the entire packet is saved by the switch, checked for CRC inaccuracies or any other problems before sending. The packet is discarded if it contains errors. The least common method is fragment-free. Its working is like that of cut-through but the first 64 bytes of the packet are stored before sending. This is because most errors and all collisions take place during the first 64 bytes of a packet.
The physical design of LAN switches differ. Shared memory is one of the three common configurations in use. This brand of switch stores all entering packets in a common buffer memory shared by all switch ports (for input and output connections), subsequently sending them out through the correct port for the required destination node. In Matrix type of switch, there is an internal grid with input ports crossing output ports. Once a packet is identified on an input port, the MAC address is matched to the lookup table to get the suitable output port. A connection on the grid where these two ports intersect is then made by the switch. The third is Bus architecture, where instead of a grid; an internal transmission path (common bus) is shared by all ports using TDMA. There is a dedicated memory buffer for each port in a switch based on this configuration, with an ASIC to control the internal bus access.
A router typically has multiple interfaces, as its main purpose is to interconnect several networks and send packets from one network to another. Every interface is a member or host belonging to a different IP network. The routing table comprises of network addresses for a router's own interfaces, which are directly connected networks, and remote networks. Though routers make their major forwarding decisions at the Network layer, router interfaces do this in Layers 1, 2, and 3. The packets at layer 1 are encapsulated into data link frames at layer 2 then encoded into bits at physical Layer 1. Router interfaces take part in processes allied to their encapsulation at layer 2. A router Ethernet interface, for example, takes part in the ARP process like other same LAN hosts. Router interfaces may vary, but Serial and FastEthernet interfaces are common.
Static and Dynamic Routes
Static routes are easily and manually configured. However, in large networks the manual operation can be quite cumbersome. Static routes have a default administrative distance of "1". If there's not a more specific match in the routing table, the default route, 0.0.0.0 network address and 0.0.0.0 subnet mask, is used to forward the packet to another router. Dynamic routing protocols require less administrative overhead. They do discovery of remote networks as they maintain up-to-date routing information, selecting the best path to target networks and have ability to find a different best path if the existing is no longer accessible or if there is a topology change. Static routing is still used nowadays, more often than not, in combination with dynamic routing.
As introduction of more industrial devices which have built-in Ethernet capabilities continues, networks keep becoming more complex and get crowded with signal traffic. This increases the necessity for Ethernet switches and routers with advanced technology which limits collisions, control bandwidth and have ability to craft virtual local area networks. Companies like Cisco, HP and IBM are racing to produce devices based on the new Energy Efficient Ethernet Standard (IEEE 802.3az). The announcement of CGR 2010 and CGS 2520 router and switch products by Cisco Systems recently designed to aid utility companies in supplying power to homes and businesses under the Smart grid project, a conglomeration of existing and technologies under development, hoped to ameliorate aging US power grid in the near future. As internet speeds get faster, through copper and fiber, switches and routers must possess features of high capacity and expandability.