It seems like only yesterday that I started writing articles on PC
networking for BE Radio. But it was actually a few years ago. At that
time, PC-based networks were not only being used for the storage of
digital audio, but also to replace the clunky mainframe systems used
for business and traffic. Today, PC networks are everywhere in the
station. Installing Ethernet or fiber cables around the station is
much easier than working with the bundles required for the previous
systems. Although most of you have become experienced and, in
general, quite proficient at installing and troubleshooting networks,
it is always beneficial to step back and review the basic theory that
makes your network run.
The name Ethernet defines an access method, or a set of rules, that
allows two or more computers to communicate over a common medium.
Other types of access methods include Token Ring and the virtually
The specific access method used by Ethernet is called carrier sense
multiple access/collision detection (CSMA/CD). In simple terms, this
means that, when one PC attached to a network is talking, the others
must be listening. If, however, two or more computers send data at
the same time, the signals will interfere with each other and no
connection will be achieved. Each PC will detect the collision and
wait for a random period of time before sending again. The process
will repeat until the destination PC successfully receives the data.
Protocols are a set of rules defining how the data will flow between
two or more computers. Depending on which network operating system
you choose, there are several possible protocols. IPX/SPX (Novell),
NetBEUI (Microsoft), NetBIOS (IBM) and DECnet (Digital Equipment) are
examples of LAN protocols. The most popular protocol currently in use
is TCP/IP, which is used exclusively to communicate over the
Internet. Protocols permit computers with different platforms to
Protocols can be classified further into one of two groups:
connection-oriented and connectionless. When two computers establish
a successful contact, the protocol takes over. A connection-oriented,
or reliable, protocol uses a three-step process: connection
establishment, data-transfer and connection release. In unreliable,
also called connectionless, protocols only the data transfer takes
place without the connect and disconnect function. Reliable protocols
ensure that the data reached the proper destination error-free.
Unreliable protocols simply send data without any checks.
The Open Standards Interconnect (OSI) model comprises a set of
protocols that attempts to define and standardize the communications
process. OSI breaks down the communications process into seven
specific layers. During the communications process, data flows from
one to the next successive layer, either from the bottom up (when
sending) or from the top down (when receiving). Figure 1 shows how
each layer depends on its surrounding layer. Each layer also has a
specific purpose. The physical layer provides the electrical and
mechanical interface to the network cabling. The data-link layer
packages the data to send over the network and disassembles data sent
to it. The network layer maintains the connection by translating
logical address information into physical addresses. It also provides
network routing and flow control across the network. The transport
layer ensures data is successfully sent and received, asking for
retransmission until it is successful. The session layer turns the
communications process on or off as needed in order to maintain a
clear communications path. The presentation layer translates data
between different computer platforms. The application layer
interfaces applications run on the computer and the network.
Other than the physical layer, the OSI is implemented through
software drivers loaded at the start-up. You should note that certain
protocols might not use all of the layers.
The topology of a network refers to the method in which the various
PCs are connected. There are three basic types of topologies: star,
bus and ring. These names describe the physical layout of the network
Most Ethernet networks use a star topology with a network hub at the
center of the star with cables run to each PC. Hubs can be cascaded
together to form other stars. Star topology allows you to centralize
the placement of network cabling for ease of troubleshooting and
The ring topology is not used with Ethernet but is used with Token
Ring-type networks. Token Ring networks are characterized by the need
to form two discreet signal paths for data flow, thus the cabling
used to wire the network will contain two discreet cables. Fiber
optic-based networks, even those used to carry Ethernet, will use a
ring topology. Fiber network cabling is sold as two-pair zip cables,
although you can purchase jacketed multipair cables.
Ring topologies are similar to star topologies in the sense that both
physically connect to a central hub-type device. Logically, however,
the ring network forms two separate data paths or rings.
Two protocols, Transport Control Protocol (TCP) and the Internet
Protocol (IP) make up TCP/IP which encompasses other protocols as
The IP delivery process provides an unreliable connection. However,
the basic function of TCP is to maintain reliable data transfer,
therefore TCP/IP is considered a reliable protocol.
Basically, data sent over a TCP/IP protocol is broken down,
sequentially numbered and encapsulated into packets that contain
specific destination address information. If all of the data packets
arrive at the specified destination, the TCP module will acknowledge
their receipt. If the packets arrive out of sequence, TCP will
attempt to put them in order. If a packet does not arrive, the
destination computer will not acknowledge receipt and the source
computer will attempt to resend.
Whether you are a certified network professional or someone who
learned it the hard way, getting back to basic networking
fundamentals always seems to help. It's time to disconnect my reliable connection.