In the assignment, you are to address some theoretical issues in the area of distributed systems or distributed computing. You will need to demonstrate your understanding of these theoretical issues or competence to apply these theoretical issues. These theoretical issues have been introduced by this unit through the textbook, weekly lectures or tutorials/labs. You will need to review these learning materials to address the following questions and prepare to do personal research if necessary.
Address the following issues that are related to replication and fault tolerance.
1. Explain the difference between the passive replication model and active replication model.
2. Explain the difference between the crash of a server and the Byzantine failure of a server.
Assume the active replication and passive replication models are available for fault tolerance.
3. If s of s+1 servers crash, explain whether the passive or active model is still fault-tolerant.
4. If s of 2s+1 servers have byzantine faults, explain whether the passive or active model is still fault-tolerant.
Recall the internet Domain Name System (DNS), which is detailed in the Week 7 unit contents and practised by Week 8 and Week 9 lab tasks, and answer the following questions.
1. Explain the hierarchical structure/arrangement of the DNS servers.
2. If the Recursive Server-Controlled Navigation is used, explain the workflow when a client sends a name resolution request.
3. If you program/implement a DNS server in Java, what multi-threading strategy will you use? Justify your answer.
Address the following issues that are related to security in distributed systems.
1. What is a public key and a private key? Given a public key, is it possible to derive/calculate its private key?
2. Assume that Alices public key is available on a web site, describe the simplest way that Bob sends a secrete message to Alice by using Alices public key.
3. There is a potential problem caused by the simple use of public key in step (2), describe the problem.
4. Describe why digital certificate can solve the problem in step (2).
Read the following scenario and address the following issues that are related to transaction and concurrency control.
Assume that objects a1, a2, and a3 are managed by a server, which provides two operations to operate the objects.
read(a): returns the value of object a
write(a, v): assigns the value v to object a
Assume that the following two concurrent transactions T and U are performed on these objects.
T: read(a2); read(a1); write(a2, a2-25); read(a3); write(a1, a1+52)
U: read(a3); read(a2); write(a2, a2+33); write(a3, a3-26)
Assume that the original values of a1, a2, and a3 are 111, 106 and 125 respectively. Answer the following questions based on the above scenario.
1. If there is no concurrency control, transactions T and U may perform the following interleaving operations on objects a1, a2 and a3. What problem can be caused by the operations? Justify your answer.
T: read(a2); U: read(a3); U: read(a2); T: read(a1); T: write(a2, a2-25); T: read (a3); U: write(a2, a2+33); T: write(a1, a1+52); U: write(a3, a3-26)
2. What requirement must be satisfied in order to avoid the problem?
3. When the above requirement in question (1) is satisfied, what would be the correct values of a1, a2 and a3 after T and U commit?
4. Give an example of possible interleaving operations that can produce the correct values of a1, a2 and a3. Note: no marks is given to this question if the operations are not interleaved.
5. The following is an example to use exclusive locks to solve the problem in question (1). Give your explanation why it can solve the problem. see image.
6. We rearrange Ts operations as follows and keep Us operations unchanged.
T: read(a2); read(a1); read (a3); write(a2, a2-25); write(a1, a1+52)
If we use the locks as below, what problem would happen? see image.
7. What will be the solution to the problem in question (6)? Give at least two different methods.