Ethernet: Key Terms & Concepts

By Bruce Cyburt, Senior Design Engineer | Wixom, MI | March 26, 2018

  • Ethernet is the IEEE 802.3 protocol standard which defines a system of interconnecting multiple devices to form a Local Area Network (LAN) for communicating and sharing information and resources.
  • HostClientComputerNodeServer, and LAN Device refer to any Ethernet device connected to a network, and any device that may have its own IP address.
  • IP refers to Internet Protocol and is the mechanism used on an Ethernet network to transport messages between nodes which include how they are addressed.
  • Gateway is a device used to regulate traffic between two different networks, while a Router is a device used to regulate traffic between similar networks. These terms are often used interchangeably for the same device and their message routing functions are often integrated into one device (Residential Gateway or Router/Gateway).
  • The IP Address is a unique 32-bit ID number temporarily assigned to each Ethernet device interconnected on an IP network. It is usually expressed in dotted decimal format as 4 groups of 8-bit integers (octets) from 0 to 255, with a decimal placed between octets (for IPv4 addresses).
  • Static IP address is an IP address assignment that is fixed inside the device itself and doesn’t change during operation. This should not be confused with addresses that act static, and instead are set by DHCP reservation in the DHCP server (which is usually integrated into the router of small networks).
  • DHCP refers to Dynamic Host Configuration Protocol which is one means of assigning IP addresses to network devices automatically (as opposed to static IP address assignment inside the device itself).
  • Subnet Mask is a second 32-bit number used to sub-divide the IP address into two or more network groups by a logical AND combination between each IP address and corresponding mask bit. The result discerns the network address/ID from its node address space. The leading bits of the network ID are flagged by set )1_ bits in the mask, and its trailing bits are clear (0) for the node ID or host address space. The maximum value of the trailing bits minus 2 determine the largest number of devices that may connect to the sub-network identified (first/zero and last/all 1’s node address numbers are always reserved).
  • IANA is an acronym for Internet Assigned Numbers Authority, a non-government, internet-specific organization that assigns and allocates IP addresses to keep the numbers unique across the globe.
  • TCP is the Transmission Control Protocol of the transport layer of an Ethernet network that makes host-host communication possible by establishing host-client connections, imposing flow-control, synchronizing sequence numbers, segmenting large amounts of data, providing error recovery and retry, and multiplexing IP addresses to specific sockets (a socket refers to a port number combined with an IP address).  TCP seeks to ensure messages are delivered and processed in the same order they are sent and uses SYNch requests & ACKnowledgement messages to establish a connection before sending data.
  • UDP or User-Datagram Protocol is another transport layer protocol that is connectionless and does not include the error recovery mechanism of TCP and does not guaranty message delivery, but is faster with less overhead as it has little error checking and no retry mechanism.
  • TCP/IP or UDP/IP Stack refers to the complete set of networking protocols required to manage communication on a network and although only its two principal protocols are commonly indicated in its acronym, the reference to stack refers to all the protocols that are required to communicate on an Ethernet network.  This stack of “software” uses the concept of groups of protocols at various layers that operate on a message that passes up/down the stack through adjacent layers with each layer doing its part for transmitting/receiving that message.  Traditionally the OSI Model is said to have a seven-layer stack, while the modern Internet reference model or TCP/IP stack has four layers. 
  • MAC address refers to the 48-bit MAChine address uniquely hard-coded into every Ethernet device.  Routers use the IP address to locate a device, but the MAC address to specifically identify the device.  In practice, you could say the IP address is used for messages sent between routers/gateways, but the IP address and MAC address is used between routers/gateways and their LAN clients.  The network router will associate the fixed device MAC address with a unique and temporary IP address that it controls, either using static IP address assignment, or via a named DHCP server that automatically make this association dynamically as the device is connected to the network.