Jackson Writes!

To print this page, please select Print from the File menu of your browser.

Network Design Basics

- - - - - - -
By Jason Jackson

The widespread use of computer networks has changed the way workers share information and communicate with one another. Once found only in large organizations, networks have become so common in small- to mid-sized businesses that few can operate efficiently without one. A company without a computer network constructs unnecessary barriers against information sharing, makes it impractical for users to share printers and other peripherals, and turns routine software updates into daunting tasks.

A computer network consists of at least two or more computers linked together by cables so they can share information, software, and peripherals. Data travels over the cables, enabling users to exchange files, access printers, and share most hardware components.

Most networks are classified by one of the following types of architecture:

Client-server network. Client-server networks designate a centralized, high-speed computer called a server as an access point for files and applications. Individual clients -- other computers connected to the network -- then retrieve information from the server. The client-server environment allows for centralized network management, and it efficiently accommodates multiple users without significantly reducing overall speed and performance throughout the network. Also, client-server networks provide enhanced access control options, as administrators can deny specific users access to particular information or resources.

Peer-to-peer network. Unlike client-server networks, where client computers retrieve information from a centralized server, each computer on a peer-to-peer network acts as a client and a server. Individual computers on a peer-to-peer network are linked directly to one another, thus avoiding the need for a dedicated server. Under this architecture, peripherals connected to a single computer become shared resources accessible from other computers on the network. However, peer-to-peer networks cannot handle multiple requests efficiently, making them good options only for very small companies with no more than eight networked computers.

In general, a network confined to a single office is referred to as a local area network (LAN). An office with multiple locations can also design a wide area network (WAN), which consists of two or more LANs linked together. Most LANs adopt a client-server architecture because of its ability to handle multiple user requests at once.

Each LAN features these essential components:

A network interface card (NIC) -- a hardware device on each computer that allows it to communicate across the network;
Network operating system software; and
Cabling or some other transmission medium, which links each machine to the network.

Numerous networking technologies have emerged over the years, but in the mid-1990s the market for business LANs mostly standardized on a type called twisted-pair Ethernet. The most common type of Ethernet had previously been 10Base-T, which transfers data at 10 megabits per second (Mbps). However, a relatively new classification called 100Base-T -- also known as Fast Ethernet -- transfers data 10 times faster than 10Base-T and is now the preferred choice for most new installations. An even faster Ethernet called Gigabit Ethernet, or 1000Base-T, can transfer data at a blistering 1 gigabit per second, but this expensive option is mostly limited to server rooms and other situations with heavy file-transfer requirements.

Networks also require several other components in order to function properly. Four of the more common elements include:

Hubs
Switches
Routers
Firewalls

Hubs

Each computer in a workgroup may be connected to a hub, a low-cost multi-port device that serves as the central location for attaching wires from individual workstations or other devices. Some, called active hubs, require a power source because they amplify the signals they receive from the network, enabling data to travel over greater distances. Others, referred to as passive hubs, serve as a conduit for the data, enabling it to travel from one device to another.

A hub looks like a thin box, lined with outlets that accommodate twisted-pair cables. Each jack corresponds to a port -- a connection point much like a telephone jack or an outlet on a power strip -- on the hub. A computer that is connected to one of these ports can share information with those connected to any of the hub's other ports. Every device that participates in a twisted-pair network requires its own port on a hub.

You'll need hubs with enough ports for every workstation, laptop, printer, or other device that connects to your network. Hubs are available that provide 2, 4, 8, 16, or more ports. If your network requires more ports than a single hub allows, you can link hubs together using crossover cables.

Switches

Workgroup computers can also be connected to a switch instead of a hub. A switch establishes a temporary connection between two ports. Switches allow a large number of devices to share a limited number of ports, because they disconnect this temporary connection once the requested information has been shared between the ports. Unlike hubs, which simply allow network traffic to flow freely across all their ports, switches pass along individual packets of data only to the port of its intended recipient.

By their design, switches tend to reduce network congestion, which can improve overall performance. The trade-off is price -- switches often cost twice as much per port as hubs. However, it is becoming feasible to forego traditional hubs entirely in favor of switches as the cost of technology declines.

Routers

A router is a device used to connect separate networks together. The most common use of routers by small businesses is to connect an office LAN to the broader Internet. However, routers can also be used to connect a LAN to a larger WAN, such as a corporate network with multiple sites.

A router determines the next network point to which it should forward a piece of information. Typically, a packet travels through several of these network points -- each equipped with a router -- before arriving at its destination. A single router can accommodate multiple users and many connections to the Internet simultaneously.

More advanced routers are often highly programmable. For example, some can be configured to allow certain types of traffic into a network, but to screen other types out. Routers increase in price depending on the number of features they support. At the high end, certain advanced routers cost tens of thousands of dollars.

Firewalls

A firewall protects a network from hackers and other unwanted outside traffic. A firewall controls access to or from a protected network by examining and evaluating traffic that passes through it. It can be a software package, a hardware device acting as a filter between your network and the outside world, or a combination of both hardware and software. There are three basic types of firewalls, with each offering different levels of security. A router can provide some degree of protection from outside attacks, but they are only suitable for smaller networks. A packet filter provides a higher level of security than a router, blocking connections from certain IP addresses it considers to be untrustworthy. Like a packet filter, an application-level proxy firewall authorizes connections, but it also provides an extra level of security be examining the incoming data stream to help detect and prevent attacks.

Originally published March 2001 on ReferralXpert.com.

- - - - - - -

Return to Feature Writing section of Jackson Writes!