Exercises – Application layer

Exercise 4.1.
With reference to the DNS hierarchy shown in Figure 4.7, divide the domain ac.nz into zones so that the number of queries required for the name resolution of ocean.physics.auckland.ac.nz by a host external to the ac.nz domain is exactly 4 and the number of zones is the minimum possible.

Figure 4.7 Examples of domain names (Exercise 4.1).

Exercise 4.5.
With reference to Example 4.2, assuming that the local name server ns.math.auckland.ac.nz has a cache and that the records relating to the previous query for the host flower.deib.polimi.it are still in the cache, determine how many queries are needed to obtain the IP address of the host tree.deib.polimi.it with recursive resolution and with iterative resolution.

Recursive resolution requires four queries as the previous resolution of the name flower.deib.polimi.it is of no use. Instead, in the case of iterative resolution, only two queries are required, as the local NS ns.math.auckland. ac.nz knows the IP address for the authoritative NS of the domain deib.polimi.it and therefore the resolution of the name takes place with direct query between two local NSs.

Figure 4.11 Recursive name resolution according to Example 4.2 (Exercise 4.5).

Exercise 4.7.
Represent graphically the exchange of messages that takes place between two MTAs using the SMTP protocol to transfer from the client ducato.com to the server fiorino.com an e-mail with the sender Andrea at the address andre@scudo.com and the receiver Mariangela at the address mari@guelfo.com. The message has as subject “Happy birthday” and contains the text lines “Happy birthday”, “from all of us”, “the cousins”.

Figure E4.1 shows the sequence of commands and responses that transfer the requested e-mail from the MTA client ducato.com to the server fiorino.com.

Figure E4.1 E-mail transfer with the SMTP protocol according to Exercise 4.7.

Exercise 4.10.
Represent graphically the access of the user Linda to an FTP server. After authenticating with the “Hello” password, Linda asks to receive the file lastnews.pdf from the directory /home/news/updates. The server replies that the requested file is not in the indicated directory. So Linda asks for the file list in the directory and discovers that the file of interest to her is called last-updates.pdf. She therefore asks for its download which is carried out successfully and leaves the FTP connection open.

As shown in Figure E4.2, Linda opens the FTP connection by providing her credentials and then the FTP client defines its port 8888 to allow the data connection to be opened for file transfer. Retrieving the lastnews.pdf file from directory /home/news/updates is unsuccessful. Nevertheless, the file last-updates.pdf is retrieved from the same directory using a new TCP data connection between server port 20 and client port 8888.

Figure E4.2 File transfer with FTP protocol according to Exercise 4.10.

Exercise 4.12.
A client accesses a server through a backbone network which provides a dedicated channel with capacity C = 100 Mbit/s and the distance along the channel between the client and the server is d = 300 km. Determine the total download time T from the server of a 1000-byte HTML page containing 8 images of 25 kbytes each by adopting non-persistent or persistent connections. It is assumed that the size of the HTTP protocol control messages is negligible and that the closure of a TCP connection can be implemented in a negligible time.

The propagation times of the messages along the channel and the transmission times of the messages are calculated first. The total propagation time of a message from client to server and vice versa is given by

   

while the transmission times of the HTML page and of each object are

   

   

The time required for the transfer of each object is in the case of non-persistent connections, while it is reduced to with persistent connections, as the TCP connection is opened only once in the latter case. Since the control messages have a negligible size, the opening of the TCP connection requires a fixed time given by RTT, indicated as and the retrieval time of each element from the server, implemented by the GET message, is simply given by the propagation time RTT, referred to as , and by the transmission time of the element itself. So in the case of HTTP protocol with non-persistent connections we obtain

   

In the case of persistent connections, it is necessary to consider that the web page and the objects are requested and retrieved with a single TCP connection and therefore similarly we obtain

   

Exercise 4.19.
In a peer-to-peer architecture of N peers with N power of 2 and upload capacity less than the download capacity identify the condition on the number of fragments k which determines a reduction by at least a factor F of the total distribution time of a file with size D bytes compared to the case of a nonfragmented file.

The number k of fragments generated from a file decomposition determines the extent of the reduction in the distribution time. Then the required condition is determined by imposing a fragmentation resulting in a distribution time F times lower than that required by a non-fragmented file, that is

   

Solving the inequality in the variable k, we obtain

   

This relationship also determines the limit on the total reduction factor in the distribution time that can be achieved with file fragmentation conseguire con la frammentazione del file

   

It is interesting to observe how the final condition found on the number k of fragments does not depend on the distribution parameters D and .