Ethernet/IP is a group of network technologies for local networks (LANs). It was introduced commercially in 1980 and was standardized in 1985 as IEEE 802.3. The standard covers specifications for several types of cables and plugs as well as the signal protocols of the OSI models. It has been possible to increase the data transfer rates from the initial 10 Mbps to 100 Gbps.
Ethernet/IP was invented between 1973 and 1974 at XEROX PARC. Digital Equipment Corporation (DEC), Intel, and Xerox specified the so-called “DIX” standard (“Digital/Intel/Xerox”) with a 10 Mbps transfer rate and a 48 bit source and destination address. This was published on September 30 1980 as “The Ethernet/IP, A Local Area Network. Data Link Layer and Physical Layer Specifications”.
The second version was published in November 1982. At the same time, efforts were made towards formal standardization.
Initially, Ethernet/IP competed against two mostly proprietary systems, Token Ring and Token Bus. Because of the better possibilities to adapt to market realities and cheaper production, at the end of the eighties, Ethernet/IP clearly established itself as the leading network technology.
Since then, the technology has developed further in order to meet the market demand for bandwidth. Ethernet/IP no longer is used only to connect computers to each other, but also for building technology and other personal devices. Up until 2010, the market for equipment using Ethernet/IP grew to over 16 billion dollars per year.
Ethernet/IP was developed further in order to support of higher bandwidths, better media playback, easier management and error processing. Individual stations communicate by sending data packets:
Data blocks which are sent or received individually. As in the case of the other IEEE 802 LAN standards, each station receives a 48-bit MAC address. This address is used to determine the source as well as the destination of each of these data packets. When a station receives a packet, it uses the destination address to determine whether this packet is relevant for it or whether it is ignored. Adapters are supplied with a globally unique address.
Originally, the standard was developed with the idea of permitting two computers to connect via a coaxial cable, where one would take the role of transmitting. This system strongly resembled the system with which radio was operated.
With a small network, this system was very reliable and simple, although it was not suitable for larger systems because of cable faults and the difficulty of finding them.
However, as each communication took place over the same cable, the information sent by one computer arrived at all computers in the network. The network card was responsible for interrupting the main processor (transmitting interrupts) only when it filtered out information that was relevant for itself.
Using a single cable also meant that the bandwidth had to be divided by the number of computers. This means that when two computers sent data at the same time, only half the bandwidth was available for each computer.
Repeaters and Hubs
For reasons of signal weakening and timing, individual data segments of the coaxial Ethernet/IP were greatly limited. Somewhat larger networks could be built with repeaters.
Earlier repeaters had only two ports, so that the sizes of the network could be at maximum, doubled. As soon as repeaters with over two ports appeared, it became possible to construct star-shaped fiber networks. First attempts were made already in 1978.
In 1989, Kalpana introduced their Etherswitches, the first switches for this technology. It worked somewhat different from bridges, where only the header was examined to decide whether a package would be sent on to another segment or would be discarded. This strongly lowered the latency and the computing power for the networks. However, one disadvantage was that damaged packages were still distributed through the network and as such, they could shut down an entire network. One countermeasure consisted of reading the complete package in a buffer (in the switch) then comparing it with a checksum and sending it on via powerful integrated circuits which are matched to the application. This way, the bridging at the hardware level is realized and the full cable speed is used.
Although Ethernet/IP networks with simple switches were a notable improvement over networks with repeaters, but they had several breaks, attacks which outwitted the switches by sending data to a machine for which they were not intended, problems of scalability / security and bandwidth bottlenecks when a lot of traffic was forced through a single connection.
Advanced network technologies in switches and routers counteracted this, e.g. the spanning-tree protocol, port security and protection characteristics such as MAC lock down, or virtual LANs, which keeps different classes of users separate, even though they are using the same physical network infrastructure.
A data packet in the cable is called a frame. A frame consists of a preamble and a start frame delimiter, followed by a header containing the source and destination MAC addresses. The middle part is composed of the data stock, including any headers for other protocols such as the Internet protocol. The frame ends with a cyclic 32-bit redundancy check in order to recognize damaged data.