Router R receives an IPv4 datagram of 3000 bytes on an incoming link. The outgoing link has an MTU of 1500 bytes. The DF bit is not set. What does Router R do?
ARouter R drops the datagram and sends an ICMP Fragmentation Needed message to the source
BRouter R queues the datagram until the outgoing link's MTU increases
CRouter R splits the datagram into fragments that each fit within 1500 bytes, giving each fragment the same Identification value but different Fragment Offset values
DRouter R forwards the oversized datagram anyway, trusting the next router to handle it
When the DF bit is clear, IPv4 routers are permitted to fragment. Each fragment gets a copy of the original IP header with the same Identification field (so the destination can group them), different Fragment Offset fields (indicating position in 8-byte units), and the More Fragments flag set on all but the last. Option A describes behavior when DF is SET. Options B and D are not valid IP behavior.
Question 2 Multiple Choice
Fragment 2 of a 4-fragment IPv4 datagram is dropped by a congested router midway to the destination. What happens at the destination?
AThe destination requests retransmission of the missing fragment from the last router that held it
BThe destination reassembles the remaining three fragments and delivers the partial data to the application
CThe destination waits for a reassembly timeout, then discards all received fragments — the entire original datagram is lost
DThe destination uses the Fragment Offset fields to reconstruct the data, filling the gap with zeros
IP reassembly is all-or-nothing. The destination collects fragments and waits for the complete set. If any fragment is missing when the reassembly timer expires (typically 60 seconds), it discards all received fragments. There is no fragment-level retransmission at the IP layer — that responsibility belongs to higher-layer protocols like TCP. This is precisely why fragmentation is expensive: losing one small fragment wastes all the bandwidth consumed by the others.
Question 3 True / False
In IPv4 networks, intermediate routers are responsible for both fragmenting oversized packets and reassembling them before forwarding.
TTrue
FFalse
Answer: False
Routers in IPv4 can fragment packets but they never reassemble them. Reassembly happens only at the final destination host. This design keeps routers stateless and fast — they do not need to buffer and track fragments across multiple flows. The cost is that any lost fragment must be handled end-to-end. If reassembly happened at routers, every router on the path would need to buffer potentially many incomplete datagrams, adding memory pressure and latency.
Question 4 True / False
IPv6 routers can fragment packets if needed, but the source host is expected to set a special flag to enable this behavior.
TTrue
FFalse
Answer: False
IPv6 routers cannot fragment packets at all. If an IPv6 packet is too large for the outgoing link, the router drops it and sends an ICMPv6 Packet Too Big message to the source. Only the source host can fragment IPv6 packets, using an extension header. This is a deliberate architectural choice that pushes complexity to the endpoints and keeps the network core fast and simple — a principle sometimes called the end-to-end argument.
Question 5 Short Answer
Why does losing a single IP fragment cause the entire original datagram to be discarded, rather than delivering the successfully received portions?
Think about your answer, then reveal below.
Model answer: IP delivers complete datagrams or nothing — partial datagrams are meaningless to higher-layer protocols. Without the missing fragment, the destination cannot reconstruct the original byte stream; the data would have a gap of unknown content. IP also has no mechanism to request retransmission of individual fragments at the network layer. The all-or-nothing design keeps IP simple and stateless; reliability is the responsibility of transport-layer protocols like TCP, which can detect the loss (via timeout or missing ACK) and retransmit the entire segment.
This also explains why fragmentation is avoided in modern networks: one dropped fragment invalidates all the bandwidth spent delivering the other fragments, making fragmentation disproportionately expensive under packet loss.