Questions: Ethernet and Frame Structure

Question 1

A switch receives an Ethernet frame destined for a MAC address it has learned. What does the switch do with the frame?

• A. Broadcasts it out all ports so every device can receive it
• B. Forwards it only to the port associated with the destination MAC address
• C. Returns an acknowledgment to the sender and buffers the frame
• D. Floods it out all ports except the source port, then learns the destination

Answer: (B)

Modern switches learn MAC-to-port associations by observing source addresses. When the destination MAC is known, the switch forwards the frame only to that specific port — creating a point-to-point link. This is a critical difference from early shared-medium Ethernet, where every device saw every frame. Broadcasting (option A) only occurs for the all-ones destination address FF:FF:FF:FF:FF:FF or for unknown destinations.

Question 2

An Ethernet frame arrives at a NIC and the computed CRC does not match the Frame Check Sequence in the frame. What happens next?

• A. The NIC requests retransmission by sending a NACK to the source
• B. The frame is corrected using the FCS as an error-correcting code
• C. The frame is silently discarded; error recovery is handled by higher-layer protocols
• D. The frame is forwarded anyway with an error flag set in the header

Answer: (C)

Ethernet performs error *detection* only, not correction or retransmission. A mismatched CRC causes the frame to be silently dropped at Layer 2. Recovery — if needed — is the responsibility of higher layers, primarily TCP, which detects missing data through sequence numbers and retransmits. This division of responsibility keeps Ethernet simple and fast while delegating reliability to layers that need it.

Question 3

Modern switched Ethernet gives each connected device its own dedicated full-bandwidth link to the switch, unlike early shared-medium Ethernet.

• T. True
• F. False

Answer: True

Early Ethernet used a shared coaxial cable where all devices competed for the medium using CSMA/CD. Modern switches replace this with separate point-to-point links: each port gets the full rated bandwidth, and frames are forwarded selectively rather than broadcast to everyone. This is why CSMA/CD collision handling is essentially irrelevant in modern networks, and why Ethernet has scaled from 10 Mbps to 400 Gbps while keeping the same frame format.

Question 4

Ethernet's Frame Check Sequence (FCS) both detects corrupted frames and triggers automatic retransmission to recover the lost data.

• T. True
• F. False

Answer: False

The FCS uses a CRC checksum to *detect* bit errors, but Ethernet does not retransmit. A frame with a bad FCS is silently dropped. Retransmission is left entirely to higher-layer protocols — TCP handles this at the transport layer through acknowledgments and timeouts. Ethernet's design philosophy is to be a fast, simple delivery mechanism; reliability and recovery belong elsewhere in the stack.

Question 5

Why does Ethernet detect errors but not correct or retransmit them? Which layer of the network stack takes responsibility for reliable delivery, and how?

Model answer: Ethernet detects errors (via the CRC/FCS) to allow corrupted frames to be discarded quickly, keeping the link layer fast and simple. Retransmission is handled by TCP at the transport layer (Layer 4), which numbers segments, expects acknowledgments, and resends anything not confirmed within a timeout window. This separation follows the end-to-end principle: reliability logic belongs at the endpoints, not in every intermediate link.

The division of responsibility is fundamental to layered network design. If Ethernet retransmitted frames, every link would have to implement complex reliability logic — even on links where data is already reliable (e.g., fiber). By delegating retransmission to TCP, the network stays efficient: UDP applications that don't need reliability avoid the overhead entirely, and TCP applications get reliability at the layer where it's actually needed.