Use this as a quick reference during interview prep. Each trade-off is a common discussion point.

---

Trade-off	Option A	Option B	When to Choose
SQL vs NoSQL	Strong consistency, joins	High throughput, flexible schema	A: finances, reporting. B: user profiles, logging
Replication: sync vs async	No data loss, higher latency	Faster writes, possible data loss	A: banking. B: social media
Normalized vs denormalized	Less storage, complex queries	Faster reads, more storage	A: normalized for writes. B: denormalized for reads
Single leader vs multi-leader	Simple consistency	Higher availability, conflict resolution	A: traditional. B: multi-region

---

Trade-off	Option A	Option B	When to Choose
Cache-aside vs Write-through	Simple, cache only hot data	Always consistent, slower writes	A: general purpose. B: critical reads
LRU vs TTL	Optimizes for access patterns	Predictable expiration	A: unpredictable access. B: time-sensitive data
Local vs Distributed cache	Faster (no network), uses RAM	Larger capacity, shared across servers	A: single server. B: multi-server

---

Trade-off	Option A	Option B	When to Choose
REST vs gRPC	Human-readable, universal, cached	Fast, typed, streaming	A: public APIs. B: internal services
Sync vs Async	Simple, immediate response	Decoupled, handles spikes	A: CRUD. B: background processing
Polling vs WebSocket	Simple, stateless	Real-time, persistent connection	A: low-frequency updates. B: real-time

---

Trade-off	Option A	Option B	When to Choose
Monolith vs Microservices	Simple to build, hard to scale	Complex, scales independently	A: small team. B: large team, high scale
Stateful vs Stateless	Server remembers context	Scales horizontally	A: simple apps. B: large-scale systems
Event-driven vs Request-driven	Loose coupling, eventual consistency	Simple, strong consistency	A: many services react. B: simple CRUD
Server vs Serverless	Predictable cost, no cold starts	Pay per use, auto-scaling	A: steady traffic. B: spiky traffic

---

Trade-off	Option A	Option B	When to Choose
Strong vs Eventual consistency	Correct, slower, lower availability	Fast, available, stale reads possible	A: banking. B: social feeds
CP vs AP (under partition)	Stop writes to stay consistent	Keep accepting writes	A: financial. B: content delivery
Active-Passive vs Active-Active	Simpler failover, standby cost	No downtime, N× servers	A: cost-sensitive. B: uptime-critical

---

Scenario	Recommended Choice
You need ACID transactions	PostgreSQL
You need high write throughput	Cassandra, Kafka
You need flexible schemas	MongoDB
You need real-time updates	WebSocket
You need to handle traffic spikes	Auto-scaling + queue
You need to protect from DDoS	CDN + rate limiting
You need to survive a zone failure	Multi-AZ deployment
You need to debug a microservice issue	Distributed tracing (Jaeger)

---

• Every choice in system design is a trade-off. Know the two options and when to pick each.
• The cheat sheet helps you quickly compare alternatives during interviews.
• Always explain why you chose one option over the other.