Miscellaneous

Sidebar 1: Ethernet Protocols

Dennis Bonotto's characterization of how network switches work surprised me. I was under the impression that network switches merely directed packets from one MAC address to another—so what's all this de-encapsulating and transforming back into bits? If you're not interested in the technical stuff, you can skip this.

As previously mentioned, a network switch is a "layer-2" device, which means that it works in the first two layers of the Open Systems Interconnection (OSI) model. In Bonotto's preferred lingo, the two layers are called "MAC" and "PHY." MAC refers to the data-link layer; it stands for Medium Access Control. PHY refers to the "Physical Layer"; Wikipedia says, "The Physical Layer is responsible for the transmission and reception of unstructured raw data between a device, such as a network interface controller [NIC], Ethernet hub, or network switch, and a physical transmission medium. It converts the digital bits into electrical, radio, or optical signals."

Bonotto wrote, in a follow-up email, "When transmitting, the MAC's role is to take the actual data and assemble full Ethernet frames/packets with address, control and the actual data as needed. However, these packets are not sent down the cable directly. Instead, they are forwarded to the PHY unit, whose role is to take those packets and 'line code' them. The reason for this is essentially to make the data more immune to noise and therefore make the whole process more reliable. On a typical 100M PHY, line coding means that the packets get converted, scrambled, and encoded into symbols for transmission, and it's these symbols that actually travel down the cable."

Wikipedia goes on to say, "common problems occurring at the Physical Layer are often related to the incorrect media termination, EMI or noise scrambling, and NICs and hubs that are misconfigured or do not work correctly."—Jim Austin