FAQ

How is latency calculated?

Answer:

Switch latency depends on how the switch handles each Ethernet frame. There are two common forwarding method:

Forwarding methodHow it worksWhere the latency comes fromPros / Cons
Cut-Through(sometimes called fast-forward)As soon as the switch reads the destination MAC address (the first 14 bytes), it starts transmitting the frame out the egress port—even while the rest of the frame is still arriving.Latency ≈ internal processing delay + time to read the first 14 bytes. This is usually a few hundred nanoseconds.✓ Lowest latency
✗ No CRC check—corrupted frames are forwarded.
Store-and-ForwardThe switch buffers the entire frame, verifies its CRC, applies filtering/QoS, and only then forwards it.Latency ≈ frame-serialization time + processing time. For a 1 500-byte frame on a 1 Gbps link, the serialization delay alone is about 12 µs.✓ Full error checking(bad CRCs are dropped)
✗ Higher latency, especially for large frames.

In short:

  • Cut-Through minimizes latency because forwarding begins almost immediately, but it may propagate errors.
  • Store-and-Forward adds the time needed to receive the whole frame, giving higher latency but better reliability.