Cut-through switching

Cut-through switching is a forwarding method in which a network switch begins sending a frame out of the destination port as soon as it has read the destination MAC address at the start of the frame, without waiting for the rest to arrive. Because the switch does not hold the whole frame, the delay it adds is small and does not grow with frame size, which suits environments where latency matters most.

The trade-off is error checking. The frame check sequence sits at the end of the frame, so a cut-through switch has already forwarded a frame before it could learn the frame was corrupt, and damaged frames travel on to the next device. Store-and-forward switching is the opposite choice: it receives the complete frame, verifies the check sequence and drops errors, at the cost of more delay. A middle option, often called fragment-free switching, reads the first 64 bytes before forwarding, which filters out most frames damaged by collisions. Switches commonly fall back to store-and-forward when the input and output ports run at different speeds.

Exam relevance: questions in this area tend to give a requirement, lowest latency or best error checking, and ask which method fits. Candidates are expected to pair cut-through with speed and store-and-forward with integrity.