Technical interview guide

CAP Theorem & Consistency Models

Why a distributed system can't have perfect consistency, availability, and partition tolerance all at once — and what real systems trade off.

Read: 23 min
Practice MCQs: 25
Interview QA: 25
Edition: v3
Editorial status: Reviewed
Relevant for: AI EngineerAgentic AI EngineerBackend EngineerBlockchain DeveloperCloud ArchitectCybersecurity EngineerData ArchitectData EngineerDevOps EngineerForward Deployed Engineer (FDE)Full-Stack DeveloperIdentity and Access Management (IAM) EngineerMachine Learning EngineerPlatform EngineerSecurity ArchitectSite Reliability EngineerSoftware EngineerSolutions ArchitectTechnical Product Manager

Scope: Vendor-neutral distributed-systems principles; current AWS architecture guidance.

Interview QA

Treat each question like a live interview question: answer out loud first (structure, assumptions, tradeoffs), then open the model answer to spot gaps and rehearse a tighter follow-up.

Curated: 2026-10-05 · Written: 2026-10-05 · Reviewed: 2026-10-03

QA-1

Explain the CAP theorem and why 'pick two' is misleading.

QA-2

Explain the consistency spectrum from strong to eventual.

QA-3

How does eventual consistency affect application design in practice?

QA-4

How would you decide which operations need strong consistency?

QA-5

Explain quorums and how they let you tune consistency.

QA-6

What should happen during a network partition, and how do you decide?

QA-7

How would you handle conflicting concurrent writes?

QA-8

Explain the trade-off between synchronous and asynchronous replication.

QA-9

What is split brain and how do you prevent it?

QA-10

How does the PACELC theorem extend the CAP theorem, and what trade-offs does it capture during normal operations?

QA-11

What causes disappearing updates across replica reads, and what architectural mechanisms guarantee read-your-writes and monotonic reads?

QA-12

What are recovery point and recovery time objectives, and how do they relate to consistency?

QA-13

How do caches interact with consistency guarantees?

QA-14

Explain why a distributed system cannot distinguish a slow node from a failed one.

QA-15

How would you design a system that must never oversell inventory?

QA-16

How does the PACELC theorem expand on CAP, and what trade-offs does it govern when the network is healthy?

QA-17

Explain read repair and anti-entropy in eventually consistent systems.

QA-18

How does consistency interact with multi-region deployment?

QA-19

What does it mean for an operation to be linearizable, and when do you need it?

QA-20

How would you test a system's consistency behaviour?

QA-21

Explain why 'the database handles consistency' is an incomplete answer.

QA-22

How would you handle a feature that needs stronger consistency than the current architecture provides?

QA-23

What is the technical difference between linearizability and sequential consistency in distributed systems?

QA-24

Explain what you would monitor to detect consistency problems.

QA-25