Recently, I was meeting with the AV staff of a local university. The objective of the conference was to introduce them to a new Ethernet based technology that is designed to carry video. I was providing a tutorial on its operation. During our exchange, I was pleased with their interest in the topic but surprised by some of their questions. They currently use an analog video distribution system and have been considering replacing it with an Ethernet based system. Their questions seemed to reveal some uncertainty about this new method. Ever since this discussion, I’ve been curious about this hesitation toward Ethernet and IP. Today, I’d like to point out that the underlying concepts of Ethernet are not new or unfamiliar. In fact, they are based on communications concepts that have been tried and used throughout the communications industry for over half of a century.
Let’s look to the past. In the early twentieth century, most data was transmitted through the air as radio signals. Next, we used the telephone network to convey data within the voice spectrum using modems. These were analog transmissions. However, in the 1960’s, AT&T Bell Labs demonstrated a digital signaling technology, called a T-carrier, that could carry both analog voice and digital data. These T-carrier circuits, which were expensive, were leased from the phone companies. Because of their cost, engineers devised a way to share circuits between different data sessions. This capability, called statistical multiplexing, or muxing for short, took advantage of the idea that not all terminals would be active at any given instant. So, data to or from one terminal could be sent while some other terminals remained idle. Both the T-carrier and statistical muxing used the concept of a data packet. That is, the digital stream was separated into blocks and sent with a small header (or metadata) appended to the front of each block. The T-1 carrier used a 193 bit block size. Statistical multiplexors used variable block sizes, depending on the service provider. The header also contained the address of the terminal to which the block was to be delivered. All of this was accomplished in the time period from 1960-1980.
Around 1975, Robert Metcalfe and his colleagues at MIT looked into combining these technologies to create the concept of a local area network. That’s how Ethernet emerged. It combined the ideas of multiplexing, sending packets, and addressing to send the data over a coaxial cable. It not only proved to be reliable and work well, but it was very fast, relative to the competing technologies such as the T-carriers. Meanwhile, satellite transmissions began to use the format called mpeg transport to carry voice, data and video. It too is also a digital stream that uses packets. It blocks the data into variable length packets. Each stream packet contains smaller packets each of which contain 4 header bytes and 184 bytes of data, audio or video.
Ethernet uses many of these older concepts. Each frame is sent by organizing up to 1500 bytes (12, 000 bits) of data, a sender’s address, a receiver’s address and a 16-bit field that provides a code to describe the payload. All of this is placed into a format that we call an Ethernet frame. This frame is sent through one or more switches. It’s a very simple technology. One of the greatest benefits of Ethernet is that the payload can be any form of media – audio, video, data, or control. In addition, the latency of Ethernet switches can be measured in microseconds. Realistically there should be minor apprehension in using this technology.