CST402 DISTRIBUTED COMPUTING

MODULE- I

PART A QUESTIONS
1. What do you mean by a distributed System?
2. List the characteristics of a Distributed System.
3. What are the various features of distributed System?
4. What are the motivation / advantages of a distributed system?
5. Discuss about the transparency requirements of distributed system.
6. List the algorithmic challenges in designing a distributed system.
7. List out few applications of distributed computing.
8. What do you mean by a distributed program?
9. Define casual precedence relation.
10. Compare and contrast physical & logical concurrency.
11. What do you mean by load balancing in a distributed environment?
Questions from model QP
1. Identify any three distributed applications and for each application, determine which all
motivating factors are important for building an application over a distributed system.
2. Assume that the surface of the past cone forms a consistent cut. Does it mean that all events
on the surface of the past cone are always concurrent? Demonstrate with the help of an
example.
3. Specify the issues in recording a global state.

TRACE KTU
PART B QUESTIONS
1. Relate a computer system with a distributed system with the aid of neat
sketches.
2. Discuss about various Primitives for distributed communication.
3. Explain in detail about the design issues of a distributed System.
4. Explain the algorithmic challenges of designing a distributed system.
5. Explain the applications of distributed computing.
6. Explain in detail about the models of distributed execution.
7. Explain the models of communication networks.
8. Discuss about the global state of distributed system.
9. Explain past cone and future cone of events in cuts of distributed computations.
Questions from model QP
1. Explain the three different models of the service provided by communication networks.
2. Explain how the causal dependency between events in distributed execution is defined
using Lamport’s happened before relationship.
3. Address the various strategies that can be adopted to satisfy the requirements of a reliable
and fault tolerant distributed system.
4. Which are the different versions of send and receive primitives for distributed
communication? Explain.
MODULE- II

PART A QUESTIONS
1. Define Logical Clock.
2. Define scalar time. Explain the rules to update scalar clock.
 3. What are the basic properties of scalar time?
4. Define Vector Time.
5. What are the basic properties of Vector time?
6. What are the applications of vector time?
7. What do you mean by a leader election algorithm?
8. What do you mean by a consistent Global State?
9. What do you mean by Termination Detection.
10. In bully algorithm if the highest node is assigned as coordinator and that site is
unreachable then an election is conducted and a new coordinator is assigned.
What happens when that highest node comes back online?
Questions from model QP
1. Define logical clock and explain the implementation of the logical clock.
2. Specify the issues in recording a global state?
3. Explain the rules used to update clocks in scalar time representation

PART B QUESTIONS
1. Discuss the implementation of logical Clock. Discuss about various Primitives for
distributed communication.
2. Explain Ring based election Algorithms in detail.
3. Explain Bully Algorithm in detail.
4. In a ring topology 7 processes are connected with different ID’s as shown: P20->P5-
>P10->P18->P3->P16->P9 If process P10 initiates election after how many message
passes will the coordinator be elected and known to all the processes. What
modification will take place to the election message as it passes through all the
processes? Calculate total number of election messages and coordinator messages.

TRACE KTU
5. Pid’s 0,4,2,1,5,6,3,7, P7 was the initial coordinator and crashed, Illustrate Bully
algorithm, if P4 initiates election, Calculate total number of election messages and
coordinator messages.
6. Explain in detail about Global state and snapshot recording algorithms.
7. Discuss in detail about Chandy Lamport algorithm.
8. Explain Termination detection using distributed snapshots.
9. Discuss the method of Termination detection by weight throwing in detail.
10. Discuss in Detail about spanning-tree-based termination detection algorithm.
11. In the ring topology shown below if process with identifier 7 initiates election, explain
the election process and show the modifications happening to the election message as
it passes through all the processes. How many election and elected messages will be
passed so that the coordinator be elected and known to all the processes?
Questions from model QP
1. Apply ring-based leader election algorithm with 10 processes in the worst-performing
case. Count the number of messages needed.
2. Apply spanning tree-based termination detection algorithm in the following scenario. The
nodes are processes 0 to 6. Leaf nodes 3, 4, 5, and 6 are each given tokens T3, T4, T5 and
T6 respectively. Leaf nodes 3, 4, 5 and 6 terminate in the order, but before terminating
node 5, it sends a message to node 1.

3. Illustrate bully algorithm for electing a new leader. Does the algorithm meet liveness and
safety conditions?
4. Clearly mentioning assumptions, explain the rules of termination detection using
distributed snapshots.
5. In Chandy-Lamport algorithm for recording global snapshots, explain how the recorded

TRACE KTU
local snapshots can be put together to create the global snapshot. Can multiple processes
initiate the algorithm concurrently?
6. Illustrate the working of spanning tree-based termination detection algorithm.

MODULE- III
PART A QUESTIONS
1. What do you mean by mutual exclusion in a distributed environment?
2. Discuss about various approaches for approaches for implementing distributed mutual
exclusion.
3. What are the requirements of mutual exclusion algorithms?
4. What are the performance evaluation metrics of a mutual exclusion algorithm?
5. List out the strategies for handling deadlocks in a distributed environment.
6. Discuss the issues in deadlock detection.
7. Describe how quorum-based mutual exclusion algorithms differ from the other
categories of mutual exclusion algorithms.
8. Calculate the rate at which a system can execute the critical section requests if the
synchronization delay and average critical section execution times are 3 and 1 second
respectively.
Rate of CS execution, System throughput = 1 / (Sync. Delay + Avg. CS execution time)
= 1 / (3 + 1) = ¼

Questions from model QP

1. What are the requirements of mutual exclusion algorithms?
2. Describe how quorum-based mutual exclusion algorithms differ from the other categories
of mutual exclusion algorithms.
3. Explain with example, how wait-for-graphs can be used in deadlock detection.

PART B QUESTIONS
1. Explain Lamports algorithm for mutual exclusion.
2. Explain Ricart–Agrawala algorithm.
3. Explain Quorum-based mutual exclusion algorithms. / Explain Maekawa’s
algorithm for mutual exclusion. / Discuss the three types of messages required for
deadlock handling in Maekawa’s algorithm. Explain how Maekawa’s algorithm
handles deadlocks
4. Illustrate Suzuki–Kasami’s broadcast algorithm.

TRACE KTU
5. Explain with example, how wait-for-graphs can be used in deadlock detection.
6. Explain / compare various models of deadlocks. / Explain any different models of
deadlock.
7. Explain in detail about deadlock handling strategies in a distributed environment.
8. Site S1, S2, S3 contains process as listed below. S1: P17, P32, P61, P72 S2: P33,
P44, P7 S3: P91, P6, P23, P28 Resources available are R1, R2, R7 All process in
site 2 is demanding for Resource R7, while R7 is assigned to the first process in site
three. Second process in site 2 is demanding resource held by third process in site
1. Evaluate the global state scenario and examine whether the resultant is in a
deadlocked condition.

Questions from model QP

1. Illustrate Suzuki- Kasami’s broadcast algorithm.

2. Compare different models of deadlocks.
3. Explain and illustrate Lamport’s mutual exclusion algorithm.
4. Discuss the three types of messages required for deadlock handling in Maekawa’s
algorithm. Explain how Maekawa’s algorithm handles deadlocks.
5. Explain and illustrate Ricart- Agrawala algorithm for achieving mutual exclusion.
6. Explain any three different models of deadlock.
 MODULE- IV
PART A QUESTIONS
1. Explain the advantages of shared memory.
2. Explain the disadvantages of shared memory.
3. What is a checkpoint?
4. What is roll back propagation?
5. Explain domino effect with an example.
6. What is global checkpoint and consistent system state.
Questions from model QP
1. Illustrate the detailed abstraction of distributed shared memory and interaction with
application processes.
2. List any three advantages of using distributed shared memory.
3. Define the no-orphans consistency condition.
4. Differentiate consistent and inconsistent states with examples.

PART B QUESTIONS
1. Explain the different ways to avoid domino effect.
2. Explain the different types of messages that occur during failure recovery.
Questions from model QP
1. Classify different log-based roll back recovery techniques.
2. What are the issues in failure recovery? Illustrate with suitable examples.
3. Show that Lamport’s Bakery algorithm for shared memory mutual exclusion,
satisfy the three requirements of critical section problem.
4. What is check point-based rollback-recovery? Explain the three
classifications of check point-based rollback-recovery.

TRACE KTU
PART A QUESTIONS
MODULE- V
1. Explain the different types of failure models.
2. Explain Byzantine agreement problem.
3. Explain consensus problem.
4. Explain interactive consistency problem.
Questions from model QP
1. Explain how consensus problem differs from the Byzantine agreement problem.
2. Define Byzantine agreement problem.
3. Differentiate between whole file serving and whole file caching in Andrew file system

PART B QUESTIONS
1. Explain the two variations of Byzantine agreement problem.
2. Explain the challenges of designing distributed file system.
3. Explain the characteristics of distributed file system.
4. Explain distributed file system requirements.
5. Explain file service architecture in detail.
6. Explain flat file service and its interface.
7. Explain the architecture of Google file system (GFS).
8. Explain the architecture of Andrew file system.
9. Explain the architecture of Sun Network file system.
Questions from model QP
1. Explain the directory service and its interface operations in a file service architecture.
2. Describe the architecture of Google file system.
3. Explain consensus algorithm for crash failures under synchronous systems.
4. Summarize distributed file system requirements.
5. Differentiate Andrew file system and NFS.
Firstly, NFS is designed to be used over a network, whereas AFS is designed to be used on
a single computer. Secondly, NFS uses a traditional file system model, with files being
stored in directories, while AFS uses a more modern "global namespace" model, where all
files are stored in a single namespace. Finally, NFS is typically used for sharing files
between Unix/Linux systems, while AFS is more often used for sharing files between
Windows and Unix/Linux systems.
6. Explain Sun NFS architecture with diagram.

TRACE KTU