Ethernet Fundamentals

EthernetEthernet was developed at Xerox PARC between 1973 and 1974, and was developed as a network technology based on a bus topology. It was originally based on the idea of computers communicating over a shared coaxial cable, and the original specification enabled them to transfer data at a rate of up to 3 Mbps. It remained a largely in-house technology within Xerox until 1979, when Xerox decided to look for partners to help promote Ethernet as an industry standard. They worked with Digital Equipment Corporation (DEC) and Intel to publish what became known as the Digital-Intel-Xerox (DIX) standard. These companies then transferred control of the Ethernet standard to the IEEE, which in turn created the now famous 802.3 committee that continues to control the Ethernet standard. The 802.3 standard initially specified communications over coaxial cable at speeds up to 10 Mbps. Since then, Ethernet has evolved from a single network technology into a standard for a family of network technologies that share the same basic bus topology, frame type, and network access methods.

Ethernet Fundamentals: CSMA/CD and MAC Addresses

Developing a networking standard at the physical level and beyond requires designing a way to send data, a way to determine which computer should use the shared cable at what time, and identify the sending and receiving computers. Ethernet deals with the first two issues by using a process called Carrier Sense Multiple Access with Collision Detection (CSMA/CD), and deals with the third issue by using data frames that contain Media Access Control (MAC) addresses to identify computers on the network. Carrier sense means that each node using the network examines the cable before sending a data frame. If another machine is using the network, then the node will detect traffic on the segment, wait a few milliseconds, and then recheck. If it detects no traffic, the node will send out a frame of data.

Multiple access means that all machines have equal access to the cable. If the line is free, any Ethernet node may begin sending a frame. It does not matter what function the node is performing. Ethernet assigns access on a first-come, first-serve basis, leaving it to other networking technologies (such as pfSense) to discriminate based on what type of traffic it is.

If two computers try to use the cable simultaneously, then a collision occurs, and both of the transmissions are lost. At the same time the network interface is sending a frame, it compares the data being sent with the data received over the cable. If there is a difference, both nodes will detect a collision and immediately stop transmitting. Then each node generates a random number to determine how long it waits to begin trying again. Whichever node generates the lowest number begins retransmission first. The losing node sees traffic on the wire and waits for the wire to be free again before attempting to retransmit.

As you can imagine, any Ethernet node will waste some time dealing with collisions instead of sending data. Moreover, as more nodes are added to the network, the collisions increase. We call a collision domain a group of nodes that hear each other’s traffic. Ethernet standards dictate that within a collision domain, there should be at most 5 segments tied together with 4 repeaters, and no more than 3 populated segments.

Ethernet requires a means of identifying sending and receiving nodes, and its method is to use special 48-bit binary addresses known as MAC addresses. MAC addresses give each network interface card (NIC) a unique address. When a computer sends out a data frame, all other NICs on the collision domain listen to the wire and examine the frame to see if it contains their MAC address. If not, they ignore the frame. If a NIC sees a frame with its MAC address, it accepts the frame and begins processing the data.

One issue with this method is that any device connected to the network cable can potentially capture any data frame transmitted across the wire. Network diagnostic programs can order a NIC to run in promiscuous mode, in which case the NIC will process all frames it sees on the cable regardless of their MAC addresses. Such programs are useful diagnostic tools but also pose a security risk, as anyone with access to the network can potentially intercept every frame on the collision domain.

External Links:

Ethernet at Wikipedia

5-4-3 Rule at Wikipedia

© 2013 David Zientara. All rights reserved. Privacy Policy