Monday, November 20, 2006

RING / FDDI / STAR topologies


RING topology


The RING network consists of a closed loop of networked devices. Each device is directly connected to another device in the loop. The RING network can cover greater distances than some other types of networks (e.g. BUS networks) due to each host’s ability to regenerate the message as it passes through it and the fact that there is no central hub requiring each device to connect to.


The most common implementation of the RING topology is the Token Ring network. These networks use a 'Token' that is passed along the RING to prevent data collision. Data can only be sent by the computer (host) with the Token, and the Token can only be held for a set period of time before being passed on to the next computer, if when the Token is received no data is required to be sent, then the Token is passed on immediately.

Original RING networks were implemented using coaxial cable limiting the maximum bandwidth achievable by the network to either 4Mbps or 16Mbps.

RING networks can be expensive to implement and costly to run – any work on the structure of the ring, or fault resulting in an effective break in the ring, results in a period of downtime (all ends of the RING network must be terminated in order for the RING to function). Also, if there is a break in any of the cables, the whole network is down. On the upside, they have the potential to deliver fantastic bandwidth over large distances.


FDDI topology

The Fiber Distributed Data Interface (FDDI) network builds on the basis of the single RING network, but consists of two RING networks running side by side. This architecture is know as 'dual-ring', and the data on each RING flows in an opposite direction (counter-rotating).


The two rings are known as ‘Primary’ and ‘Secondary’. The Secondary ring is generally used as a back-up RING, with the Primary delivering up to 100Mbps capacity. In some circumstances where the Secondary ring is not being used for backup, the capacity can be doubled and therefore effectively deliver 200Mbps of bandwidth.

An advantage of FDDI over a standard RING network is the fact that FDDI uses fibre optics, and so is capable of far higher bandwidth and further distances between hosts (and so overall distance).


Also, by using two rings instead of one, FDDI allows for an element of redundancy that a RING network cannot. If a computer is taken out of the network, the secondary ring forms a loop using the remaining computers and thus maintains the network between the remaining computers.


STAR topology

The STAR network is built using a central 'connection' known as the hub, off which all other network devices are connected. This 'hub' can be a hub, switch or router, and depending on the device, ranges from a simple data link to actively processing which messages are sent to which machine.

The basic hub design consists of messages being sent from a single computer to the hub, and the hub then forward the message onto all other computers connected to the hub.


STAR networks have a typical bandwidth of 100Mbps, though this does vary, and use Ethernet cabling (CAT 5).

Data collision can be high on a STAR network, and this increases as the network grows. Whenever there is a data collision, the whole network is 'frozen' for an instance and then effectively 'reset'. As data collision increases as more and more devices are added to the network, this can have a major impact on performance.

Devices such as switches and routers reduce data collision, as these actively manage data direction, effectively reducing the traffic on the network and therefore reducing the possibility of data collision.


Comparison of STAR and RING topologies

Below is a table showing comparison of RING and STAR networks:


RING

STAR

Reliability

The RING network is limited by design, as the whole network is down whenever there is a fault in the line

Good, the network remains up, even if there is a failure taking out a computer.

Cost

RING networks are reasonably easy to implement, as there is no central hub.

A central hub is require and cabling needs to be routed from all attached devices to this hub.

Speed

The RING network can offer high speeds on reasonable size networks as data only goes in one direction. Though there on larger networks speed becomes an issue do to data having to pass through every computer between the sender and receiver.


Can offer excellent speed, even on larger networks if combined will switches and routers. Data is only sent from host to receiver, with few other devices involved.

Maintainability

Network has to be down to add or remove a computer from the network. Though only cabling is required between to a computer on the network to the new computer

Very good as new computers can be added to the network without any down time. Though if new cabling is required this can prove to be an issue, as it has to be routed to the nearest hub.

Expandability

There is a limit to the number of computers that can be included on a RING network before there is a drop in performance.

There is good potential for expansion on a STAR network, as with the use of switches and routers, additional network activity is minimal.

Data Collision

Data Collision is not an issue on a RING network, as only the computer that currently holds the token can transmit data.

Can become an issue when there is a large amount of network activity either on large networks or those that the functionality of switches and routers.

Saturday, November 18, 2006

The CSMA/CD Protocol

Each PC connected to an Ethernet network connects using a Network Interface Card (NIC), which controls data connectivity between the host PC and the network. Note: Each NIC has a unique identifier called the MAC address (which can be seen as the ‘Physical Address’ when using the ‘IPCONFIG /ALL command in MSDOS).

The Carrier Sense Multiple Access / Collision Detection protocol (CSMA/CD) means that all NICs connected to the network must listen and wait until there is no traffic on the network.


At this point each NIC has an equal chance to transmit data across the network. If any of the other NICs are transmitting data across the network, there will be a signal called a 'carrier' on the network that informs other NICs that the network is busy. This is defined as 'Carrier Sense'.

When there is no traffic on the network, each NICs has an equal chance of transmitting data, and this is defined as 'Multiple Access'.

As data can taken a defined period of time to travel across the network, and isn't instant (though to us it may appear to be so), there is a chance that two NICs may transmit data at the same time.


When this is detected by the nearest NIC at the point where the two transmissions meet, a signal is sent across the network that causes all transmissions to cease for a period of time before transmissions resume.

This is defined as 'Collision Detection', and when this happens each NIC resets an internal random delay that avoids the same collision occurring when transmissions can resume.

The CSMA/CD protocol by design allows an equal chance for each NIC to transmit across the network allowing the network to be shared by several NIC connected devices.