Cloud Computing MCQs - Cloud Storage, Scalability and Availability (Set 5)

Q1. Which storage type keeps data as items with unique keys and metadata in a flat namespace and is accessed mainly through HTTP APIs?

a) Block storage

b) File storage

c) Object storage

d) Tape storage

Q2. A relational database on a virtual machine needs a low-latency raw volume that it formats with its own file system. Which storage type fits best?

a) Block storage

b) Object storage

c) Archive storage

d) A content delivery network

Q3. Several VMs need simultaneous read-write access to the same hierarchical directory tree over NFS. Which storage type is most appropriate?

a) Object storage

b) A block volume attached to one VM

c) Network file storage

d) Ephemeral instance storage

Q4. Adding more identical instances behind a load balancer to handle more traffic is an example of

a) Vertical scaling

b) Horizontal scaling

c) Scaling up

d) Over-provisioning

Q5. Elasticity differs from plain scalability because elasticity emphasises

a) Automatically adding and removing resources to track demand closely over time

b) Only permanently adding hardware

c) Using faster CPUs

d) Increasing network bandwidth

Q6. A request must pass through a web tier and a database tier, each independently 99.9% available. What is the end-to-end availability?

a) 99.90%

b) 99.80%

c) 99.99%

d) 99.95%

Q7. Two redundant servers, each 99% available, are placed in parallel so the service works if at least one is up. What is the combined availability?

a) 99%

b) 99.9%

c) 99.99%

d) 98.01%

Q8. An SLA guarantees 99.9% monthly uptime. What is the maximum allowed downtime in a 30-day month?

a) 4.32 minutes

b) 8.76 hours

c) 52.6 minutes

d) 43.2 minutes

Q9. A component has a mean time between failures of 990 hours and a mean time to repair of 10 hours. What is its availability?

a) 99.9%

b) 98.99%

c) 90%

d) 99%

Q10. According to the CAP theorem, when a network partition occurs a distributed data store must trade off between

a) Consistency and availability

b) Consistency and durability

c) Availability and latency

d) Partition tolerance and scalability

Q11. A database keeps accepting writes on both sides of a network partition and reconciles conflicting replicas afterwards. It is best classified as

a) CP and strongly consistent

b) AP and eventually consistent

c) CA with no partition tolerance

d) Serialisable ACID across all replicas

Q12. Which load-balancing algorithm sends each new request to the backend with the fewest active connections?

a) Round robin

b) IP hash

c) Random selection

d) Least connections

Q13. To route /api/* requests to one server pool and /images/* to another, you need a

a) Layer 4 load balancer

b) Layer 7 (application) load balancer

c) Plain DNS A record

d) ARP proxy

Q14. Servers A, B and C have weights 3, 2 and 1 in weighted round robin. Out of 600 requests, how many does server A receive?

a) 100

b) 200

c) 300

d) 600

Q15. An internal reliability target such as '99.95% of requests succeed over a 28-day window' is called an

a) SLI

b) SLO

c) SLA

d) MTTR

Q16. What is the purpose of a cooldown period in an auto-scaling policy?

a) To prevent repeated scaling actions before metrics stabilise after a previous action

b) To lower the physical temperature of CPUs

c) To delay billing for new instances

d) To shut down idle regions

Q17. Which load balancer feature sends all requests from the same client session to the same backend instance?

a) Health checks

b) Connection draining

c) Session affinity (sticky sessions)

d) TLS offloading

Q18. What is the primary way a content delivery network improves performance for users across the world?

a) Increasing CPU on the origin server

b) Compressing database indexes

c) Replicating block volumes within one zone

d) Caching content at edge locations closer to users