Problem set 2. (1.-3.10.2002)

1. Suppose the SACK-option (Selective ACK) is in use. Show with a diagram how the segments are sent in the following situations.
   1. The size of the congestion window is 8 MSS. At first the sender succeeds to send one segment correctly, but after that an error burst corrups the following 3 packets.
   2. The size of the congestion window is 8 MSS. The second segment is routed to a congested router and it arrives to the receiver first after the fourth segment sent. Other segments arrive in order.

2. Suppose that the SACK option is not used, but instead the partial acknowlegdements of NewReno. How does the TCP handle the situation where after a succesfull transmit of one segment, three following segments will get lost. After that there are no problems in the transmission. The congestion window is 16 MSS.

3. SACK and the NewReno partial acknowledgements both try to solve a problem in TCP congestion control. What is the problem they try to solve? Compare these solutions. What are the advantages and disadvantages of these solutions?

4. RED buffer and ECN
   1. How does a RED buffer work? What are its benefits?
   2. A TCP implementation wants to use ECN (Explicit Congestion Notification). What extra functionalities are needed in TCP and IP protocols? How is the ECN usage negociated? Suppose a segment on a TCP connection using ECN ends up into congestion. How does the information about this reach the sender and how does the sender behave after that.

5. IP protocol and large packets
   1. How large amounts of data can be sent using IP protocols? What factors constrain the size of IP datagrams?
   2. Is it possible when using IPv4 protocol to sent bigger than normal messages? For example can a supercomputer A send a message of 264144 bytes to another supercomputer B using IPv4 protocol? If it can how is it done? What happens when this superdatagram comes into a network that cannot transfer such big datagrams?
   3. How does IPv6 handle large packets?

6. IPv6 includes the flow label field.
   1. Give examples of situations where flow labels are useful. What kind of sendings do not benefit from flow labels? Would it be useful to use flow labels when transmitting large amounts of data in a TCP connection?
   2. What happens when a router receives an IP packet with a flow label unknown to that router so that the router does not know how to handle the packet? Can this cause problems?