ICT 427 DISTRIBUTED COMPUTING
FINAL EXAM
Distributed Systems Architecture Consensus Middleware and Fault
Tolerance Questions and Answers Verified Solutions 2026 2027
1. What fundamental characteristic distinguishes a distributed system from a centralized
system?
A. Execution on a single multi-core processor
B. Autonomous components communicating over a network appearing as a single
coherent system
C. Shared central memory accessible by all processing nodes
D. Dependence on a single master clock for synchronization
Rationale: A distributed system consists of independent computers linked through a network that
coordinate actions using message passing to present a unified system to end users.
2. According to the CAP Theorem, which two properties can a distributed data store
simultaneously guarantee during a network partition?
A. Consistency, Availability, and Partition Tolerance simultaneously
B. Either Consistency and Partition Tolerance (CP) or Availability and Partition
Tolerance (AP)
C. Consistency and Availability only, while ignoring network partitions
D. Parallelism and Fault Tolerance exclusively
Rationale: The CAP theorem proves that in the presence of a network partition (P), a distributed system
must choose between guaranteeing strict consistency (CP) or high availability (AP).
3. What clock synchronization algorithm relies on a time server to estimate network
latency using round-trip time measurements?
A. Cristian's Algorithm
B. Lamport Timestamps
C. Vector Clocks
D. Paxos Consensus
,Rationale: Cristian's algorithm synchronizes client clocks with a time server by measuring round-trip
transmission time and setting client time to server time plus half the delay.
4. What fundamental limitation of Lamport Timestamps is addressed by using Vector
Clocks?
A. High communication overhead during message passing
B. Inability to determine causality between arbitrary events when timestamp values
differ
C. Dependency on physical hardware atomic clocks
D. Inability to order events occurring on a single node
Rationale: Lamport timestamps provide a total ordering of events but cannot determine whether two
events are causally related or concurrent; vector clocks track complete causal histories.
5. What RPC mechanism handles marshaling and unmarshaling of parameters on the
client side before network transmission?
A. Client Stub
B. Server Skeleton
C. IDL Compiler
D. Transport Layer Handler
Rationale: The client stub packages local procedure parameters into a standard network format
(marshaling) and forwards them across the network layer.
6. Which consensus protocol algorithm guarantees safety under asynchronous network
conditions with up to f crash failures using a leader-driven model?
A. Two-Phase Commit (2PC)
B. Raft Consensus Algorithm
C. Chandy-Lamport Algorithm
D. Bully Algorithm
Rationale: Raft manages replicated log consensus through explicit leader election, log replication, and
safety constraints while tolerating non-Byzantine node crashes.
7. What primary drawback affects the classical Two-Phase Commit (2PC) protocol in
distributed databases?
, A. Blocking behavior if the coordinator crashes during the commit phase
B. High likelihood of cascading dirty reads across worker nodes
C. Inability to achieve atomic transaction logging
D. Excessive disk storage consumption on participant nodes
Rationale: 2PC is a blocking protocol; if the transaction coordinator fails after participants vote,
participating nodes remain blocked holding locks indefinitely.
8. What distributed snapshot algorithm captures a consistent global state without
freezing system execution?
A. Berkeley Algorithm
B. Chandy-Lamport Algorithm
C. Ricart-Agrawala Algorithm
D. Raymond's Tree Algorithm
Rationale: The Chandy-Lamport algorithm uses marker messages traveling along communication
channels to record local process states and channel messages consistently.
9. Which election algorithm assumes that every node knows the process ID of all other
nodes in the distributed network and elects the node with the highest ID?
A. Bully Algorithm
B. Ring Algorithm
C. Leased Leader Algorithm
D. Zab Protocol
Rationale: In the Bully Algorithm, when a process detects leader failure, it initiates an election by sending
messages to higher ID nodes, "bullying" lower ID processes.
10. What mutual exclusion algorithm uses logical timestamps to order requests and
requires permission from all other processes before entering a critical section?
A. Suzuki-Kasami Algorithm
B. Ricart-Agrawala Algorithm
C. Maekawa's Algorithm
D. Token Ring Algorithm
Rationale: Ricart-Agrawala optimizes Lamport's algorithm by requiring processes to send timestamped
requests to all nodes, entering critical sections only after receiving all permissions.
FINAL EXAM
Distributed Systems Architecture Consensus Middleware and Fault
Tolerance Questions and Answers Verified Solutions 2026 2027
1. What fundamental characteristic distinguishes a distributed system from a centralized
system?
A. Execution on a single multi-core processor
B. Autonomous components communicating over a network appearing as a single
coherent system
C. Shared central memory accessible by all processing nodes
D. Dependence on a single master clock for synchronization
Rationale: A distributed system consists of independent computers linked through a network that
coordinate actions using message passing to present a unified system to end users.
2. According to the CAP Theorem, which two properties can a distributed data store
simultaneously guarantee during a network partition?
A. Consistency, Availability, and Partition Tolerance simultaneously
B. Either Consistency and Partition Tolerance (CP) or Availability and Partition
Tolerance (AP)
C. Consistency and Availability only, while ignoring network partitions
D. Parallelism and Fault Tolerance exclusively
Rationale: The CAP theorem proves that in the presence of a network partition (P), a distributed system
must choose between guaranteeing strict consistency (CP) or high availability (AP).
3. What clock synchronization algorithm relies on a time server to estimate network
latency using round-trip time measurements?
A. Cristian's Algorithm
B. Lamport Timestamps
C. Vector Clocks
D. Paxos Consensus
,Rationale: Cristian's algorithm synchronizes client clocks with a time server by measuring round-trip
transmission time and setting client time to server time plus half the delay.
4. What fundamental limitation of Lamport Timestamps is addressed by using Vector
Clocks?
A. High communication overhead during message passing
B. Inability to determine causality between arbitrary events when timestamp values
differ
C. Dependency on physical hardware atomic clocks
D. Inability to order events occurring on a single node
Rationale: Lamport timestamps provide a total ordering of events but cannot determine whether two
events are causally related or concurrent; vector clocks track complete causal histories.
5. What RPC mechanism handles marshaling and unmarshaling of parameters on the
client side before network transmission?
A. Client Stub
B. Server Skeleton
C. IDL Compiler
D. Transport Layer Handler
Rationale: The client stub packages local procedure parameters into a standard network format
(marshaling) and forwards them across the network layer.
6. Which consensus protocol algorithm guarantees safety under asynchronous network
conditions with up to f crash failures using a leader-driven model?
A. Two-Phase Commit (2PC)
B. Raft Consensus Algorithm
C. Chandy-Lamport Algorithm
D. Bully Algorithm
Rationale: Raft manages replicated log consensus through explicit leader election, log replication, and
safety constraints while tolerating non-Byzantine node crashes.
7. What primary drawback affects the classical Two-Phase Commit (2PC) protocol in
distributed databases?
, A. Blocking behavior if the coordinator crashes during the commit phase
B. High likelihood of cascading dirty reads across worker nodes
C. Inability to achieve atomic transaction logging
D. Excessive disk storage consumption on participant nodes
Rationale: 2PC is a blocking protocol; if the transaction coordinator fails after participants vote,
participating nodes remain blocked holding locks indefinitely.
8. What distributed snapshot algorithm captures a consistent global state without
freezing system execution?
A. Berkeley Algorithm
B. Chandy-Lamport Algorithm
C. Ricart-Agrawala Algorithm
D. Raymond's Tree Algorithm
Rationale: The Chandy-Lamport algorithm uses marker messages traveling along communication
channels to record local process states and channel messages consistently.
9. Which election algorithm assumes that every node knows the process ID of all other
nodes in the distributed network and elects the node with the highest ID?
A. Bully Algorithm
B. Ring Algorithm
C. Leased Leader Algorithm
D. Zab Protocol
Rationale: In the Bully Algorithm, when a process detects leader failure, it initiates an election by sending
messages to higher ID nodes, "bullying" lower ID processes.
10. What mutual exclusion algorithm uses logical timestamps to order requests and
requires permission from all other processes before entering a critical section?
A. Suzuki-Kasami Algorithm
B. Ricart-Agrawala Algorithm
C. Maekawa's Algorithm
D. Token Ring Algorithm
Rationale: Ricart-Agrawala optimizes Lamport's algorithm by requiring processes to send timestamped
requests to all nodes, entering critical sections only after receiving all permissions.
