Networking Topologies and TCP/IP Protocol
- Length: 1371 words (3.9 double-spaced pages)
- Rating: Excellent
It just doesn't get much simpler than the physical bus topology when it comes to connecting nodes on a Local Area Network (LAN). The most common implementation of a linear bus topology is IEEE 802.3 Ethernet. All devices in a bus topology are connected to a single cable called the bus, backbone, or ether. The transmission medium has a physical beginning and an end. All connections must be terminated with a resistor to keep data transmissions from being mistaken as network traffic. The terminating resistor must match the impedance of the cable.
One advantage of bus topology is that small networks are fairly easy to set up and does not require specialized networking equipment. It is also fairly inexpensive to set up, since it requires the least amount of cable and equipment. Adding or removing nodes is fairly easy, but moving nodes without affecting neighboring nodes can be difficult.
Troubleshooting media problems on a bus networks can be very tedious, since a break in the backbone will bring down the entire LAN. For this reason, bus topology is not considered one of the more robust network topologies, compared with star or mesh. A loose or missing terminating resistor can also bring down a LAN.
In this topology, all nodes are connected to a central device, usually a hub or a switch. Each connected device has a dedicated, point-to-point connection between the device and the hub. The star network topology is by far the most widely implemented topology in use today.
Star topology networks require more cabling than bus, but the tradeoff comes in the form of a more solid network topology. A break in the network media will only affect a single node, since every node has a dedicated connection to the central device; a hub or switch. This also makes the central device a Single Point of Failure (SPOF). Redundant or failover switches are often used to eliminate the SPOF in a star LAN.
Building a star topology is much more expensive and time consuming than the bus network. The increased costs come in the form of cabling and the central device(s). However, star topologies can be easily expanded to accommodate more nodes and troubleshooting is much easier because connectivity problems are simpler to isolate than a bus network.
Perhaps the most redundant, fault-tolerant of all network topologies is the mesh LAN. Each node is connected to every other node for a true point-to-point connection between every device on the network.
Should one cable fail, network traffic can be rerouted to the destination via an alternate path.
While the mesh topology is by far the most robust of the LAN topologies, the tradeoff is cost, complexity, and difficulty of troubleshooting. Mesh requires the most cabling of any topology. Because of these disadvantages, a true mesh topology is seldom used. Instead, a hybrid mesh, with redundant cabling paths to critical network devices like routers, is sometimes used in the core layer of networks.
Ring topology is actually a logical ring, which means that while the data travels in a circular pattern from one node to another, the cabling pattern can take on any form. Since each station acts as a repeater, greater distances can be spanned with a ring network than other physical topologies. Ring networks can also offer higher bandwidth with almost no signal degeneration. Fiber-optic media is commonly used in modern ring topology.
An advantage of the ring network is troubleshooting and isolating a break in the media is much easier than other topologies. Some ring technologies have the ability to locate the break for the technician. When a break does occur, the network is effectively down. Many modern rings are built with a redundant or secondary ring to make a more fault-tolerant network. This dual-ring is referred to as Fiber Distributed Data Interface (FDDI).
Comparison of Network Transport Protocols
Ethernet (IEEE 802.3 CSMA/CD) has several advantages over other transport protocols like Token Ring and wireless. First of all, Ethernet hardware and cabling is relatively inexpensive, compared with FDDI or wireless. The simplicity of Ethernet also reduces the expense of maintenance and support. Second, Ethernet allows for various types of networking media such as twisted pair, coaxial, and even fiber. Ethernet also offers more security than wireless, since connectivity requires physical access to the wire.
While wireless is often considered Ethernet, it uses CSMA/CA (Carrier Sensing Media Access/Collision Avoidance). Wireless network security and signal interference has been a real concern for network administrators. Most of these concerns have been addressed, but the reputation of being a largely free and wide-open media still lingers.
Perhaps the greatest advantage of wireless networks is the freedom from physical cabling. Laptop users can finally use their machines as they were intended; away from a desk. Unfortunately, wireless LAN standards (IEEE 802.11) are considerably slower than their wired counterparts. 802.11a wireless is capable of speeds up to 72Mbps; however, it is currently restricted to a throughput of 54Mbps by the FCC.
While Token Ring (IEEE 802.5) bandwidth was speedy for 1980's (4 or 16Mbps), its performance would be hardly sufficient for all but the smallest of LAN networks today. What really made Token Ring such an attractive network transport protocol for its time was its method for handling collisions, called token passing. A station must posses the token to transmit data. When a host has data to transmit, it waits until the token is passed, changes one bit on the token and passes the data. The receiving station must wait until the token is passed and remove the data and place the empty token back on the ring. While collisions are expected and even anticipated in Ethernet, in Token Ring networks, they are almost non-existent. Token Ring even has error corrections methods to detect if a token is hung up by clearing the ring and starting a new token.
Unlike IEEE 802.5 Token Ring, FDDI uses a dual-ring technology for redundancy and fault tolerance. Should the primary ring fail, the secondary ring can take over with no disruption to users on the network. FDDI uses multiple tokens, which is vastly different than 802.5 Token Ring. Since FDDI used fiber and a media, collisions are virtually impossible. Fiber-optic pulses can only travel in one direction, so separate cables must be used to transmit and receive.
FDDI is undoubtedly the most expensive, fastest, and the most secure way to transmit data across a network. Fiber-optic cable is immune copper and wireless problems like Electromagnetic Interference (EMI), crosstalk, attenuation, and degradation of the signal. The expense of operating FDDI comes not only from the fiber-optic cabling itself, but the specialty networking equipment required to send and receive the light pulses.
ISO OSI vs. TCP/IP model
The ISO OSI, seven-layer networking model is a conceptual model only. Not one working protocol truly matches up layer-for-layer with the OSI model. The designers of the OSI model never intended it to be a working protocol, but rather as a reference tool.
The TCP/IP model was designed to allow communication among a variety of multi-vendor, independent systems.
There are versions of the TCP/IP model ranging from as few as three layers to as many as five. The top three layers of the OSI model (Application, Presentation, and Session) are combined into the Application layer of the TCP/IP model. The transport layer is the same in both models. The Internet layer of the TCP/IP model equates to the Network layer of the OSI model. Both the Data Link and Physical layers of the OSI model map to the Network Access layer of the four-layer TCP/IP model. One similarity of both models is that they assume packet-switched technology, not circuit-switched technology (SONET/ATM) is in use.
Harwood, Mike, Tittel, Ed (Ed.) (2003). Network+ Exam Cram (Exam N10-002). Indianapolis, IN: Que Certification.
Kozierok, Charles M. (2004). The TCP/IP Guide. Retrieved August 14, 2005 from http://www.tcpipguide.com/index.htm.
McQuerry, Steve (Ed.) (2004). CCNA Self-Study: Introduction to Cisco Networking Technologies (INTRO). Indianapolis, IN: Cisco Press.
Tomsho, Greg, Tittel, Ed, Johnson, Greg (2004). Guide to Networking Essentials [University of Phoenix Custom Edition e-text]. Thomson Learning, Inc. Retrieved Aug 14, 2005 from University of Phoenix, Resource, NTC/360 Network and Telecommunications Concepts Web site: https://ecampus.phoenix.edu/secure/resource/resource.asp.