This section briefly addresses a few issues you should consider before projecting.

MENU

Networking & Windows 9x

The choice of media is primarily determined by four major factors.
The first is what exists in the environment today. Is there existing wiring that could be reused?
The cost of "pulling" wire can be a major component of larger network installations.
The second is how "noisy" is the environment?
"Noisy" refers to the amount of electromagnetic energy that exists in the environment.
Most office environments are considered fairly "quiet"; on the other hand, a manufacturing floor with motors and compressors starting and stopping at random times is considered quite noisy.
Third, the topology to be used has a minor implication. Finally, the number of workstations to be used does play a factor.

ThinNet
Thin Coax (also called ThinNet) is the simplest medium to use if the number of workstations is small, 10 or fewer.

Thin Coax uses easily available coax cable and BNC connectors.
Each machine has a "tap" into the ThinNet via its BNC T-connector.
To ensure the needed electrical "load" exists, a 75-ohm resistance cap is used to terminate the coax at each end.
The major disadvantage is that a fault in the cable, the T-connectors, or the resistance caps will affect all machines on the network.
A break in the coax does not produce two networks that cannot communicate.
Both fragments fail (preventing any network communications) because the "load" on the coax is no longer balanced.
Depending on the number of workstations it is possible the wire itself will generate enough load for the network to continue functioning, but this is not guaranteed.
Also, ThinNet is only appropriate for Ethernet topologies.
It is not possible to use this technology with a token ring network.

ThinNet
Advantages Disadvantages

High transmission rates supported (10 MBPS)the network. Limited distance between farthest points.
Low maintenance--there are no active components in the network.
Single failure in the coax will cripple the entire network.
Simple to install--only have to install the cable.
Expense of installation does not scale-in small installations, installation is quite inexpensive, provided the cable is left on top of the floor.
As installation grows, however, installation can be quite expensive, especially if the cable is to be pulled through crawl spaces. Cable is thicker than UTP.
Also, some zoning requires that Teflon jacket cables be used to avoid exposure to dangerous gases in the event of fire.
Easy to add a new machine--just connect the new tap into the network, and connect tap to the BNC connector.
Low security--very easy to add a new machine to the network.
Done quickly enough, the network software may not be aware of the new machine in the network.

Unshielded Twisted Pair (UTP)
Unshielded Twisted Pair (also called UTP) is a very well-understood technology.
It is based on common telephone wiring technology, so there is a wealth of technology surrounding UTP from multiple vendors.

Unlike ThinNet, however, there are additional components that must be purchased to make a UTP network active.
These additional pieces (hubs, concentrators, and so on.) do add to the expense of the installation; however, they can also be used to isolate cabling failures that would cause the collapse of a ThinNet network.
UTP can be used by either Ethernet topologies or token ring technologies.

Unshielded Twisted Pair
Advantages Disadvantages

Well understood technology with multiple vendors for each additional component.
Requires additional hardware components such as hubs and concentrators.
Affects of failure localised--single failure only affects the machines connected to the failed component.
Example: failure in cable monitored periodically. between hub and machine only affects the single machine.
Very inexpensive media--the UTP wire and connectors are very inexpensive.
The medium is very susceptible to noise interference.
Ability to build hierarchical networks on the
Initial installations and small installations are media level to further isolate traffic and failures. more expensive than ThinNet because of the additional cost of hubs and concentrators.
Scales well into larger networking environments.
Easy to add a new machine--simply connect a new machine to nearest hub.
Low security--very easy to add a new machine to the network. Unlike ThinNet however, some hubs and concentrators can alert a centralised manager of the change in configuration.

Wireless
The newest medium available in the local area network market is the wireless LANs, there is no wire connecting the machine to the network. In fact there is no wire anywhere in the network.

There are two technologies currently being developed. First is infrared (IR) based systems.
Network traffic is transmitted as an infrared signal to a repeater usually located on a wall.
The other technology is radio frequency transmissions, usually low-power FM.
The technology is relatively new to the LAN market.
The bandwidth is still relatively low (1 MBPS as compared to UTP at 10 MBPS).
In addition, the installation problems can be every bit as complicated as a wire based system.
Some systems require direct line of sight between the machine and the repeater to be able to participate in the network.

Wireless
Advantages Disadvantages

No wires direct to each machine--useful in generally open environments or in older buildings, where cable access is not possible.
Very special hardware at each machine to communicate with the centralised repeater.
Much lower bandwidth than UTP or ThinNet.

Some systems require direct line of sight with the nearest repeater.
Local failures are usually local--failure of the local transceiver does not usually affect others on the network.
Local failure could cause complete network failure--just as with ThinNet network, a machine that is jabbering (constantly transmitting) can keep other machines from participating in the network.
Additionally, central repeater failure can cause complete failure.
Not affected by electromagnetic noise--useful in manufacturing environments. (water absorbs IR).
In addition, both IR and FM can leak beyond your environment and could be picked up by unauthorised personnel.
IR is less susceptible than FM.
IR is most appropriate as long as the environment does not have excessive moisture
Very simple to move machines--simply move Moving may require additional tools or expertise
Easy to add a new machine--simply add a new machine. Very low security--very easy to add a new machine.

Ethernet
(also called CSMA/CD, or media backbone) was one of the original topologies used in local area networks.

Every machine connected to a single network hears all traffic on the network.
Though it hears all of the conversations on the network, it only listens to the traffic that is addressed to it or is broadcast to a group, or to everyone on the network.
The technology has a very strict etiquette in the way it communicates.
When a machine has something to transmit, it listens to the wire.
If someone is already transmitting, it will wait. If there is no one transmitting, it will begin transmission.
Transmission of the signal from one end of the network to the other is not instantaneous.
There is a propagation delay.
The implication is that a machine may begin transmitting because it did not hear the transmission of another because it has not yet reached its location in the network.
The chances of this increases as the number of machines on the network or their transmission needs increase (load on the network, usually measured in percentage of total bandwidth).
When two or more machines are transmitting at the same time, a collision will occur.
This destroys the messages being transmitted.
When this is detected (usually by the machine nearest the collision), the entire network is shut down, and the machines transmitting wait a random time before they attempt to retransmit their message.
Ethernet networks are not limited to one massive backbone network.
Many different devices (known as bridges and routers) are available on the market to help localise the network traffic and reduce the amount of network traffic on each segment.
Though this sounds fairly complicated, it is by far the most popular of the two topologies.
It is the most proven technology, and it is available on many different types of machines and supported by many different vendors.

Ethernet
Advantages Disadvantages
Most popular topology in use today--very high level of vendor support.
Typically designed to use no more than 30% of total bandwidth to avoid collisions and collision recovery.
Capable of using many different media solutions - can be implemented on ThinNet, UTP, or wireless.
Scales very well--capable of designing very large networks while limiting the amount of data being transmitted.
Requires additional support from additional hardware and protocol to support bridging and routing techniques.
Requires great insight into the traffic patterns on the network to predict the affect of adding additional machines to the network.

Token Ring
Token ring technology was originally developed by IBM in the early 1980's.
Unlike Ethernet, where all machines are logically connected to the same section of connected media, in token ring, each machine is connected to the two machines adjacent to it.

These machines are known as the nearest upstream neighbour and nearest downstream neighbour.
Thus machine are connected in a daisy chain fashion to form a logical ring. It is called a logical ring, because the physical topology is actually star topology, where the concentrator is at the centre and each machine is at one of the points.
Around the ring races a token.
The token is a specially formatted frame that grants the holder of the token the right to transmit its information on the network.
This is very similar to the token
system used by railroad systems to permit multiple trains to use a single piece of track in either direction.
As with the Ethernet topology, each machine in a network hears each frame transmitted.
Unlike Ethernet, a machine retransmits each frame it receives (unless it was the original transmitter of the frame).
Because of the token, there is no collision as in Ethernet, and so it should be possible to use a greater percentage of the bandwidth, as well as make a more accurate estimation of the impact of adding additional machines to the network.
Token ring networks are not without their disadvantages, however.
First, the propagation delay is greater than in Ethernet topology (because each machine is basically retranslating the frame).
This retransmission does provide one beneficial service in that the signal that forms the frame on the wire is fresh at each machine.
Any loss of signal strength (attenuation) is removed as it is retransmitted.
This is not without its problems, however.
There is a potential timing problem known as jitter, which can cause the data to be corrupted if the internal clocks on the sender and receiver are slightly out of sync.
Just as Ethernet, token ring has a number of automatic recovery mechanisms.
Unlike Ethernet there are more potential failures and methods of recovery.
For example, if there is a break in the ring, the system will automatically attempt to discover the point of the failure and create a loop back path, which will allow communications to continue.
There is also, the jabber situation.
Unlike Ethernet, token ring can detect this and attempt to disconnect the jabberer from the network.


Token Ring
Advantages Disadvantages

Newer technology--avoids some of the problems Less support from third-party providers.
Still fairly proprietary to IBM, even it is an IEEE standard (802.5).
Token ring hardware is usually more expensive than Ethernet hardware.
Scales very well--easier to predict the affect of adding machines to the network.
1. Use if Token ring is already installed in your environment.
A direct connection to an SNA based mainframe via a Token ring Interface Connector (TIC) is needed.

NetBEUI
NetBEUI is an extension to the original protocol designed by IBM for its networking solutions in the mid-1980's.

It is characterised by a very small protocol header (extra bytes added to the data to instruct the protocol on the meaning of the data).
NetBEUI's protocol header is only 14 bytes.
This helps increase the speed of the protocol because less data is being transmitted, and less is needed to be analysed by the protocol software (also called a stack).
Unfortunately, this small header does not possess enough information to permit NetBEUI to be routed (used by routers to help reduce traffic).
This severely limits its ability to be used in any but the smallest installations.
It should be noted, however, that IBM has attempted to address the routing issues with token ring.
Token ring has information in the Media Access Control (MAC) header (before the NetBEUI header) known as the source routing information.
This information can be used to route NetBEUI, provided you use source routing routers.
Source routing is not as robust or full -featured as routing techniques used by other protocols.

Limited media support to UTP.

NetBEUI
Advantages Disadvantages
Fastest protocol option.
Not appropriate to use in large installations, unless using a token ring.

TCP/IP
TCP/IP was one of the original data communications protocols.

It began as a Wide Area Network (WAN) protocol and has since become very popular as a LAN protocol, since it allows a single protocol to be used for both local area and wide area connectivity.
In addition to this, there is a very rich set of utilities and services in common access for TCP/IP.
Most of these utilities have already been ported to the Windows NT platform.
Finally, the programming interfaces to the network via TCP/IP are well-established and available (Sockets and RFC/NetBIOS).
TCP/IP's protocol header is bigger than NetBEUI's, 64 characters.
However, the information in the header allows TCP/IP frames to be routed (because the ultimate destination is in the frame).
TCP/IP is typically slower then NetBEUI (under LAN Manager, by 40%), but the ability to build very large hierarchical networks using routers is typically viewed as worth it.

TCP/IP
Advantages Disadvantages:
Routable Protocol
Requires careful planning and management of the IP address space.

SPX/IPX
SPX/IPX (usually simply referred as IPX) is a derivative of the oldest networking protocol, XNS.

This protocol is tightly associated with Novell NetWare. In a Windows NT and Windows for Workgroups environment, the ri networking services (LAN Manager based services) can use the IPX protocol as its native protocol by using the NetBIOS/IPX protocol drivers provided.
This is a great advantage if the environment is a mixture of Novell NetWare and Windows NT, since you can continue to have a single protocol running on your network.

SPX/IPX
Advantages Disadvantages:

Routable Protocol More overhead than NetBEUI; however, less of a performance penalty than TCP/IP.
Does not support the same level of routability controls (via the LMHOSTS file), which TCP/IP does.
Allows single protocol in mixed NetWare / Windows NT environments

Networking & Windows 9x