0

I am trying to understand what an Ethernet Switch internally considers a single port.

My current understanding is that when a seller advertises an Ethernet Network Switch, like the one below, has five ports, they actually mean that you may connect to it five standard UTP cables that each hold four unshielded twisted pairs.

enter image description here

Moreover, I understand that Ethernet specification doesn't assume four UTPs (a single common UTP cable) but actually works on a single unshielded twisted pair.

Finally, knowing that switches forward frames to a MAC address accessible via a given port I am wondering if that means that there are actually four ports for each cable in the switching table (i.e. each unshielded twisted pair has a separate port in the table) or that a group of four pairs (a cable) is considered one port in the table.

Improve this question
2
  • "actually works on a single unshielded twisted pair." No. Some ethernet standards require two twisted pairs, some ethernet standards require four twisted pairs, and some ethernet standards are on various optical media. The cabling categories are defined by ANSI/TIA, while the cable classes are defined by ISO/IEC. It is the IEEE that sets the ethernet standards on the various media set by the cabling standard bearers. ANSI/TIA defines Category-5 with four twisted pairs, which has applications other than ethernet, and IEEE uses it for some ethernet standards.
    – Ron Maupin ♦
    Commented Jun 5, 2022 at 15:18
  • You are trying to think of it as one automaker (IEEE) defining the roadways (government), but it is really the other way around. The roadways (various UTP standards, various optical media, etc.) are built for various types of traffic, and the automakers are required to build vehicles (ethernet, POTS, token ring, overhead paging, serial communications, etc.) for the roadways.
    – Ron Maupin ♦
    Commented Jun 5, 2022 at 15:25

1 Answer 1

Reset to default
4

Each port allows you to connect a single device to the switch, using a point-to-point connection. That device may also be another switch though. Accordingly, switches are required when connecting more than two devices to each other.

I understand that Ethernet specification doesn't assume four UTPs (a single common UTP cable) but actually works on a single unshielded twisted pair.

That is not correct. Each port requires a category 5 cable, each containing four twisted pairs. Categories aren't defined by IEEE 802.3 but by ISO/IEC 11801 and Ethernet just uses those.

The lower-speed standards 10BASE-T (requiring only category 3) and 100BASE-TX just use two pairs from the four, so a sub-standard cable may also work. 1000BASE-T and faster require all four pairs to work.

I am wondering if that means that there are actually four ports for each cable in the switching table (i.e. each unshielded twisted pair has a separate port in the table) or that a group of four pairs (a cable) is considered one port in the table.

The latter. Each connection with four twisted pairs forms an atomic port, capable of a single link. If any used pair fails the link goes down.

While 10BASE-T and 100BASE-TX use one pair for transmitting and another for receiving (dual simplex), all faster TP standards split the data stream into four equal lanes and transmit those in parallel. All pairs are used in both directions simultaneously (full duplex), using hybrids and echo cancellation with adative equalization.

Improve this answer

Not the answer you're looking for? Browse other questions tagged or ask your own question.