Ch. 5 Transport Layer

8. What is a socket?

A network IP address and a port address together constitute a communication socket. Applications can send data over the network using a socket. Each network connection is identified by a pair of sockets - the port and IP addresses of the sender are the socket at one end of the connection and the port and IP addresses of the receiver are the socket at the other end of the connection. Together, the two sockets make up the network connection.

9. What is a port?

A port identifies a communication channel on the host. Each application on the PC is assigned a separate port by the client operating system.

Why is it necessary to define port numbers at the receiving end for network services (such as web, email, etc.)?

Before sending a datagram, the sender needs to know the port number at which the receiver is running the server application. Defining standard ports makes it easy for senders to guess the correct port number from the name of the service. If standard port numbers were not defined, senders would have to first ask receivers for the correct port number before sending any application data.

11. What is connection establishment in TCP? Why is it necessary?

Connection establishment relates to the initialization of sequence numbers for a TCP connection and exchange of this initialization information between sender and receiver. Connection establishment is necessary to help the sender and receiver to generate suitable sequence numbers and to ensure that both sender and receiver know what sequence numbers to expect from the other side.

12. Describe the three-way handshake used in TCP.

In the 3-way handshake, the sender and receiver exchange a total of 3 packets using which they inform each other of the initial sequence number they intend to use for the connection. In the first packet, the sender sends its ISN. In the second packet, the receiver sends its own ISN as well as an acknowledgment of the sender's ISN. In the third packet, the sender acknowledges the receiver's ISN

13. What are some application scenarios where UDP may be more useful than TCP? Why?

Real time applications such as voice and video commonly use UDP. Also, most network services such as DNS use UDP

4. What is segmentation?

Segmentation is the process of breaking down large blocks of data into smaller blocks. This is done because the maximum packet size at the network layer is 65,536 bytes. If an application has a bigger block of data to transfer (audio and video files come to mind), some entity needs to chop the larger block of data into smaller segments before handing the segments to the network layer. Therefore the transport layer performs segmentation when necessary.

5. Why is segmentation useful?

Segmentation is useful because if the transport layer did not perform segmentation, the application developers would have to do it while developing their applications, and also keep track of lost and duplicate packets.

7. How does TCP provide reliability?

TCP provides reliability through the use of sequence numbers. Each TCP datagram is assigned a sequence number. The receiver keeps track of the sequence numbers of the incoming datagrams and alerts the sender if a datagram is not received within a reasonable amount of time. The receiver also periodically sends acknowledgments of incoming datagrams to the sender. If an acknowledgment is not received within a reasonable amount of time, the sender assumes the worst and resends the datagram.

3. Why are two protocols, TCP and UDP, defined at the transport layer instead of just TCP?

TCP provides reliable transmission of data, but at the cost of computation overhead and network delays. Many applications, particularly voice and video applications prefer real time delivery more than reliability. For these applications a simpler Transport layer protocol called the User Datagram Protocol has been defined. UDP eliminates all the computational processing associated with TCP

2. What are the functions of the transport layer?

The main function of the transport layer is segmentation and re-assembly of user data. If requested, the transport layer keeps track of these segments to account for packet loss or duplication during transmission by the network layer. The Transport layer also provides a mechanism (port addresses) to distinguish between the data streams created by each individual application on the computer. Finally, the Transport layer provides flow control by allowing receiving computers to specify the data transfer rate.

6. What is reliability in the context of the transport layer?

The network layer only provides best effort delivery and may drop packets or duplicate packets during transmission. Some entity needs to resolve these losses and duplications. Reliability in the context of the transport layer therefore refers to the correction of lost and duplicate packets.