webpointmorpheus Network Info
Network Topology & Hardware

     Overview      Network Topology      Network Hardware      5-4-3 Rule
     Token Ring      ARCNET      AppleTalk-FDDI-ATM      Notes

©2005 - material compiled by Bob Carnaghi, www.webpointmorpheus.com

Overview     Top of Page

Topology with regards to computer networks is the physical layout of the machinery, as well as the logical function of their interconnection. While this distinction may not make a clear first impression, it becomes more important as networks become increasingly complex. For example, it is possible to have a star type physical topology that behaves like a bus topology in a logical sense. Please keep in mind the following basic distinctions:

• physical topology = layout of computers, hardware, and wiring.
• logical topology = behavior of electronic circuitry between the computers, hardware, and wiring.

If you have a difficult time with the acronymn drenched terms used in this document, check out the Network Definitions Page.

Topology     Top of Page

There are basically five types of physical topology:

1. Star - connections to a central point.
2. Bus - connections along a linear bus-like structure.
3. Ring - a circular arrangement.
4. Mesh - a hybrid between star and ring.
5. Wireless - no physical connections. See the Wireless Networking Page.

Topologies are rarely implemented in a pure sense of their definition. Typically they are physically one type and behave logically as another. For example, a network may connect as a star to a central hub, but the wiring inside the hub performs the function of a bus. This is called a star-bus topology.

   Star     Top of Page

Star topology is by far the most widely used. The Star topology works well in a small to medium range network, such as an office or a building. Nodes are easy to add, as all that is required is to run a single wire to the new node and connect it to the central hub or router. This arrangement makes the installation cost per node cheaper. In the star topology the central hub or router is the common link as well as the potential bottleneck. The reliability of a star type network is directly related to the central point of connection, be it a hub, switch, or router.

   Bus     Top of Page

A Bus topology is similar to a straight pipeline for connecting computers to pass data streams between them. This can be an extremely effective topology for adding extra nodes and reliability. However, this topology fails at longer lengths, and if there is a break anywhere in the bus connection, all systems downstream are disconnected. This system does not work well with Fiber Optics.

   Ring     Top of Page

Ring topology was successfully pioneered in the early days of networking by IBM, and is still in wide use today. Token Ring networks have the advantage of easy expansion, as it is easy to add additional nodes. However, if there is a break in the network cable, the system may shut down.

   Mesh     Top of Page

A true Mesh topology has a link from any single computer to any other computer. This is typically a fault-tolerant system wherein any break in the data transfer can be circumvented due to the other routes available. Mesh topology works well on a very small scale (less than about 10 computers) or on a grand scale like the Internet.

Network Topology
Topology	Speed	Signal/Cable	Distance	Nodes	Spacing	Cost	Notes
10Base2	10 Mbps	Baseband RG-58 Coax	185 Meters/segment	30 Nodes/segment	0.5 Meters per node	Low	Thinnet - good for small size networks. Moderate shielding from EMI.
10Base5	10 Mbps	Baseband RG-8 Coax	500 Meters/segment	100/segment	2.5 meters per node	High cost	Thicknet - cost high per foot due to cabling. Good shielding from EMI.
10BaseT	10 Mbps	Baseband RJ45 UTP	100 Meters hub-node	1024a Nodes/hub	NA	Low	Common - good for medium size networks.
Token Ring	4/16 Mbps	Can use STP for increased distance	Uses MAU - Multiple Access Unit				IBM Proprietary IEEE 802.5
10BaseFL	10 Mbps	Baseband Fiber SC or ST	100 Meters hub-node	1024a Nodes/hub		High	Fiber Opticb cabling system that has been outdated by faster technology.
100BaseFX	100 Mbps	Baseband Fiber SC or ST	400 Meters hub-node	1024a Nodes/hub		High	Fiber Opticb cabling system that improves on 10BaseFL
100BaseT	100BaseTX	Cat5 RJ45	Uses 2 pairs of the UTP				Consider for upgrades of existing horizontal cabling
	100BaseT4	Cat3 RJ45	Uses all 4 pairs of the UTP				Consider for upgrades of existing horizontal cabling
	100BaseVG	Cat3	Not a true Ethernet	Controlled by 802.12 Standards			Also called 100BaseVgAnyLan
Gigabit Series 1000BaseTc	1000BaseCX	Baseband Fiber SC or ST	25 Meters	Uses a special 150 ohm coax cable			Governed by IEEE 802.3z standard
	1000BaseSX	Baseband Fiber SC or ST	500 Meters				Governed by IEEE 802.3z standard
	1000BaseLX	Baseband Fiber SC or ST	5-70 Kmd				Governed by IEEE 802.3z standard

 

Network Hardware     Top of Page

Listed below are the common hardware elements that are used in Network architecture. Note that these items are physical items that are installed to create an interconnected network of computers. These are the elements that pass data which has been formatted at various layers of the OSI model to other portions of the network (LAN, WAN, etc.) Click the link to see the detailed information for the network hardware component.

• NIC - Layers 1 & 2, Physical Layer and Data Link Layer
• Repeater - Layer 1 - the Physical Layer
• Hub - Layer 1 - Physical Layer
• Bridge - Layer 2 - the Data Link Layer
• Router - Layer 3 - the Network Layer
• Switch - Layer 2 - the Data Link Layer
• Modem - Layer 1 - the Physical Layer

    NIC - Network Interface Card     Back to Hardware List

A NIC is a hardware device that is physically installed into a computer which serves as the connection between the computer and some sort of network of computers. The function of a NIC is to organize data handed down from the computer into frames which are passed along to the network, to check incoming and outgoing frames for errors, and to track receipt of incoming frames. The NIC works on Layer 1 (Physical Layer) to transmit data over the wire, fiber optic cable, or Radio Frequency. The software of the NIC works at Layer 2 (Data Link Layer) to format and error-check the data frames.

A NIC is configured to work specifically with certain IEEE defined standards, such as Ethernet, Token Ring, etc. The NIC will have a unique MAC address that is 'burned' in to the card for unique identification on the network. An additional IP address is assigned to the NIC to identify the host in its connection to the other LANs or to the Internet.

NICs can have one of five basic bus standards for internal connection to the PC: PCI, ISA, EISA, MCA, or PCMCIA. PCMCIA bus types are used in some desktop and all laptop computers. The limitations of data transfer of the NIC are determined by the bus type of the NIC. A NIC will have a BNC, RJ-45, or some combination of these types of connectors for interfacing to the wiring of the network. Fiber Optical connector types will be found on NICs capable of FDDI.

NICs are access protocol specific, and must be configured with a unique IRQ, (IRQ 10 by convention,) I/O port, (0300 and 031F by convention,) and base memory address (D8000 by convention,) for the computer on which they are installed. They are typically PnP capable, and if installed into an OS with PnP capability require little or no configuration. Manual configuration of a NIC may require software configuration or manual positioning of jumpers or DIP switches.

    Repeater     Back to Hardware List

A repeater is also known as an active hub, see below.

    Hub     Back to Hardware List

A hub is a network device that connects computers to a centralized point. Typically, a hub receives packets from one of the computers that are connected to it, and broadcasts those packets to all the computers that are connected to it, even the computer that sent the packets. Hubs use a physical star topology, and can vary in their logical topology. Ethernet hubs use a logical bus topology, and Token Ring hubs use a logical ring topology for network connectivity.

Hubs can be passive or active or switching. A passive hub gets its power from the computers that connect to it, and does no formatting, error checking, or refinement to the data packets it receives. It simply passes along exactly what it receives. A patch panel is an example of a passive hub. An active hub requires an external power source, and may refine the quality of the data packets it receives. Some active hubs support remote administration and configuration. A switching hub allows any port on the hub to be connected to any other port on the hub without interruption. This increases data throughput, and can be used to create separate collision domains.

Hubs can have from 4 to 32 ports. They can be stacked by using crossover cables or uplink ports to add nodes to a network. Most hubs use RJ-45 connectors.

    Bridge     Back to Hardware List

The main purpose of a bridge is to connect networks that use different access protocols. Connections among networks with diverse protocols such as Ethernet, Token Ring, and Wireless are possible with a bridge. A bridge can also segment a large network that uses a similar protocol to reduce traffic and congestion. A bridge works at the Data Link Layer of the OSI Model.

There are three basic modes of bridge operation.

• A transparent bridge stores a list of addresses of all the nodes on a network. Traffic on the network is then forwarded based upon the entries in the table of MAC addresses the bridge has accumulated.
• A source routing bridge passes frames of data based upon the destination routes of nodes on the network that it has retrieved and stored.
• A translation bridge converts from one network protocol to another.

    Router     Back to Hardware List

The primary function of a router is to connect different networks, typically a LAN to the Internet. Routers direct incoming network protocol packets from one network to another based on OSI network layer info that is stored in the incoming packets. A router can determine if a data frame contains a MAC address on its internal network, or an IP address from outside its network, then forward the data accordingly. Additionally, routers have the capabliity to determine the status of a frame using CRC (Cyclic Redundancy Check) and to discard the frame if it's faulty. Routers operate at the Network Layer of the OSI Model.

Routers store path information in router tables. They do not forward broadcast requests to other networks, the broadcast request packets are discarded by the router.

Routers are capable of using several protocols to fulfill their purpose of filtering and forwarding data. A list is below.

• BGP - Border Gateway Protocol. An routing protocol for interconnection to the Internet.
• RIP - Routing Information Protocol. An interior routing protocol for the LAN.
• OSPF - Open Shortest Path First. An interior routing protocol for the LAN.
• IGRP - Interior Gateway Routing Protocol. An interior routing protocol for the LAN.
• EIGRP - Enhanced Interior Gateway Routing Protocol. An interior routing protocol for the LAN.

    Switch     Back to Hardware List

A switch provides a virtually dedicated and direct connection from one node to another on the network. When a series of data packets are sent over a network through a hub, they are broadcast to all nodes on the network. Only the intended node receives the data addressed to it, all others reject the data. With a switch, software connects two nodes directly for the duration of the transmission. Due to this capability, a switch is unique in that it can offer full bandwidth connection status between the communicating nodes. There is a virtual dedicated connection between the sending and receiving nodes through the switch. This serves to reduce network traffic and decrease data transfer rate.

A switch can typically run in one of two modes. In cut-through mode, the data transmission is passed withoug caching. In this type of transmission the first data packets often arrive at their destination before the last data packets are sent. In store-and-forward mode, the switch stores and buffers the entire transmission before processing and forwarding. The transmission may be further processed during this type of transmisson.

    Modem     Back to Hardware List

MOdulator/DEM - a computer hardware element used to carry out data communications via telephone lines. The primary function of a modem is to convert binary data bits to analog information (for transfer over the telephone line) and from analog to binary (when received from the phone line) to feed into the computer system. modems have three operating modes:

1. Simplex mode: A modem used to send signals in only one direction.
2. Half-duplex mode: A modem that can communicate in both directions, (send and receive), but in only one direction at a time.
3. Full-duplex mode: A modem that can communicate in both directions simultaneously.

Modems can be either internal to the computer system, typically installed as an expansion card, or external to the system. The use of modems has begun to be surpassed by the growth of high-speed Internet connections such as DSL and cable. However, for the average Internet user, the dial-up connection fostered by a modem remains a bargain.

Baud rate is a definition of telephone data transfer which has been handed down to computing standards through the use of modems. Originally, modems were rated equally to phone transmission, which is in Baud. Baud rate is described in terms of data symbols, which is a small group of bits. Baud has a maximum rate limit of 2400. When modems exceeded this transfer rate, their data transfer rate started to be described in multiples of Baud.

The 5-4-3 Rule^e     Top of Page

The distances between nodes in a complex network can make collsion detection impossible. When sending data packets, each node on the network must be able to determine if its packets are colliding with the data packets of another node. If the last packet of a node's transmission leaves the node before the first packet arrives at the destination, there will be a situation where a collision goes undetected. The 5-4-3 rule allows for a working equation to limit the size of a network that counts the segments, hubs, and repeaters.

The 5-4-3 Rule states that in a collision domain, no two nodes may be separated by more than:

• 5 segments
• 4 repeaters
• 3 populated segments

Note that when using the 5-4-3 Rule, a hub counts a both a repeater and a segment.

Token Ring     Top of Page

Token Ring is an older topology that was pioneered by IBM and is still in use in their systems. Typically all the computers in a Token Ring are connected in a ring-like fashion. The topology is noted for its use of 'token passing' to all the nodes on the ring. A node can only transfer the data when it receives the token, which signals that the ring is clear for data transfer. The nodes are connected through an MAU (Multiple Access Unit) which is similar to a hub, but is proprietary to the Token Ring topology. The standard for these units is IEEE 802.5.

ARCNET     Top of Page

ARCNET is an antiquated topology (early 80's) that may (does) still exist in small pockets today (as of this writing.) ARCNET works in a true star topology, and later installations used UTP cable. ARCNET typically uses a passive hub, and can have a length of up to 600 meters per segment. For this reason, it was attractive at the time, and is most likely the reason that ARCNET installations can still be found. ARCNET was eclipsed by Ethernet and Token Ring topologies.

AppleTalk, FDDI, & ATM     Top of Page

Listed below are three technologies that are in use today but may not be encountered on a regular basis. While LocalTalk, an older Apple protocol, is on the decline, FDDI (Fiber Distributed Data Interface) is on the increase. ATM (Asynchronous Transfer Mode) is used in broadband digital signal data transfer.

    AppleTalk

AppleTalk is the current protocol suite of the Apple Computing industry, and LocalTalk is the antiquated protocol that is at the root of the AppleTalk Protocol Suite. AppleTalk is a routable protocol that uses a simple 24 bit network address, thus:

     12.6

The first portion of the address is the network number, and the second is the node ID. The AppleTalk frames must be encapsulated into IP frames for transport. The protocol suite consists of the following protocols:

• NBP - Name Binding Protocol. Maps computer and server names to addresses similar to that listed above.
• ADSP - AppleTalk Data Stream Protocol. A Session Layer protocol similar to TCP.
• DDP - Datagram Delivery Protocol. Network Layer protocol similar to IP.
• ATP - AppleTalk Transaction Protocol. Used in a similar manner to TCP to transport instructions/packets.

    FDDI

Fiber Optic cabling offers the best medium for speed and quantity of data transfer. This technology is the most expensive (as of this writing) due to the price of connectors, cable installation requirements, and labor intensive splicing, cost of network cards and components, etc. The speed of data transfer has reached 200,000 Mbps with bandwidths to 1,000 GHz in laboratory environments. The price of Fiber Optic cabling has become more reasonable in the last few years. Typically, the cabling is used selectively for longer runs such as between buildings. The length of a Fiber Optic segment may reach 2 kilometers or more. In order to more fully understand the intricacies of Fiber Optic cabling, an understanding of the properties of light waves/particles is helpful.

    ATM

ATM a cell relay technology that operates at the first two levels of the OSI model: Physical and Data Link layers. ATM technology has come about as the needs for greater data transfers in speed and quantity have been demanded. ATM is a standard that is implemented in ISDN lines and is the only medium that is capable of being scalable for voice, data, and video technologies. In essence, ATM is capable of rates from Ethernet standards up through T3 (45 Mbps.)

Notes     Top of Page

1. The 1024 node limit is rarely or never reached due to pragmatic limitations, such as Operating Systems, physical location limitations, etc.
2. Fiber Optic cabling generally has the capabilies of long transmission - up to 2 km, no EMI, high cost, and is difficult to install.
3. Capable of Full Duplex, which means data transmission in both directions at the same time.
4. Yes, that's Kilometers. This standard uses lasers to shoot the signal down the cable. The greater distances use repeaters as well.
5. When using a Switch, the 5-4-3 Rule is considered Null & Void.