Global Sources
EE Times-Asia
Stay in touch with EE Times Asia
EE Times-Asia > Networks

Choosing the right topology for Ethernet

Posted: 16 Jun 2003 ?? ?Print Version ?Bookmark and Share

Keywords:ethernet? topology? bus topology? star topology? network?

As Ethernet's importance grows, cable engineers must understand its various topologies and their pros and cons. The design guidelines for the different Ethernet types must be examined, and the common problems that occur during installation and operation reviewed.

Ethernet operates in a variety of physical structures called topologies. Two of the more commonly used topologies are bus and star, each having advantages and disadvantages in addition to cost and reliability trade-offs.

The bus topology is the simplest of the two. Each station on the network is linked to a central cable connected to all other devices on the network. Because of this configuration, each station listens to the traffic from all other stations. Although this is relatively simple to build and operate, it may run into performance problems quickly as the traffic load rises. Because every network interface card (NIC) must use the same media as all the other NICs, collisions are common and network performance drops exponentially as traffic volume rises and retransmissions become necessary. Bus topologies include 10Base2 and 10Base5.

One way to solve the performance problems associated with media sharing is to construct the network in a star configuration. This is the most common topology in today's Ethernet-based LANs, and is usually constructed using shielded twisted pair or unshielded twisted pair (UTP) cabling. The star configuration has the advantage as each NIC has a dedicated physical medium to communicate to the network equipment. This has allowed the bottleneck to move from the physical media to the first component of the network, the hub.

LAN evolution

The hub has the same problem of performance vs. traffic load, as it is also a shared resource and the packets are subject to collisions. The ability to incorporate additional hubs and segment the network has benefited users.

In essence, the need to segment the network became crucial as traffic loads increased. The switch solved this problem by creating multiple dedicated segments, which were used to combine the traffic from the hubs back to the main servers. However, as technology improved and costs reduced, the hubs were eliminated completely and every NIC was connected to a switch port.

All of these enhancements require a speed of more than 10Mbps to send all of the combined traffic to the servers. So 100BaseT became a popular backbone technology for the hub architecture and early switches. As the cost per port decreased and port density increased, 100Mbps moved to the desktop and Gigabit Ethernet became the technology of choice for backbone. Although GbE (1000BaseT) can be implemented over UTP cabling, it is most commonly executed using fiber optics in the backbone as 1000BaseSX (850nm) or 1000BaseLX (1,300nm).

Today's typical network has 10BaseT or 100BaseT on the desktop and 1000BaseT on the backbone. New technology is currently available to move the desktop to speeds of 1Gbps over UTP and the backbone to 10Gbps on fiber. However, the LAN deployment of these new technologies is not happening at a rapid pace. The reason for this deferral is that the average user does not have LAN bandwidth problems.

The user has access to the local resources needed without any significant bottlenecks, which have moved again and are now outside of the LAN, on the WAN. If a user has a dedicated 10Mbps or 100Mbps network connection and then shares the connection to the Internet with everyone else in the company, you can quickly understand where the next bottleneck exists.

The cost of high-speed Internet connectivity makes it unrealistic for most companies to implement this type of performance to the outside world. The good news is that most applications that operate over these connections do not need the same level of performance as local resources.

Test tools have progressed as networks have continued to develop. Although there are more parameters and standards to understand, test gear simplifies this for the user and adds control so that testers can be configured by predefined profiles created by project managers or engineering consultants.

- Chuck Ganimian

Product Manager, FrameScope/WireScope

Agilent Technologies Inc.

Article Comments - Choosing the right topology for Ethe...
*? You can enter [0] more charecters.
*Verify code:


Visit Asia Webinars to learn about the latest in technology and get practical design tips.

Back to Top