How explain the basis for determining the minimum frame size in ethernet?

Answer

Size is 64 bytes.
Destination Address (6 bytes)
Source Address (6 bytes)
Frame Type (2 bytes)
Data (46 bytes)
CRC Checksum (4 bytes)

46 bytes must be transmitted at a minumum, with additional pad bytes added to meet frame requirements.

There is a fundamental minimum frame size imposed by the headers, but that isn't it.

Think back to the days when ethernet was a multi-drop fat yellow coax, rather than the mesh of switched point to point cat5 links it is today.

A sender needed to send the packet and determine whether or not it got received by others on the wire. Reasons why it might not get received were mostly down to collisions with other transmitters transmitting on the same wire at the same time.

To determine if there was a collision, the sender listened to his own transmission to see if it could be recieved. If it was garbled due to a collision, it would know. Hence the name CSMA/CD (Carrier Sense Multiple Access with *Collision Detect*).

However transit time down the wire isn't instantaneous. If you could send a tiny packet at the same time as someone at the far end of the wire, a receiver in the middle of the wire might receive both signals at the same time, garbled, yet they pass over each other and get received by the far ends just fine. So they think the guy in the middle got it, but it didn't, thus breaking the Ethernet model on which 802.1D relies.

To prevent this, you want the head of the packet to transit from one end of the wire and back again before the tail of the packet finished transmission. If there's a transmission anywhere on the wire overlapping, you'll hear it collide at your receiver.

The maximum transmission time is lengthened by repeaters. You need to account for the time to transit through the maximum number of wires and repeaters on a LAN segment.

The header length issue is a red herring. The transit time is independent of bit rate. So as bit rates go up, the header length goes down. So trying to use header size to determine that the packet is long enough is not safe. Safe at 10MBPS might not be safe at 100MBPS, hence the need for a time based packet length minimum.

In short, the packet has to be long enough for the collision detect part of the CSMA/CD algorithm to work.

First answer by ID984492084. Last edit by ID984492084. Question popularity: 13 [recommend question]