Compare

Core Concepts

• Availability

  Single instance vs Redundant instances (N+1)

• CAP Theorem

  CP (Consistency + Partition tolerance) vs AP (Availability + Partition tolerance)

• Latency vs Throughput

  Process immediately vs Batch before processing

• Scalability

  Vertical scaling vs Horizontal scaling

Storage & Data

• Caching

  TTL expiry vs Explicit invalidation on write

• Database Indexes

  No index vs Index

• Database Replication

  Synchronous replication vs Asynchronous replication

• Database Sharding

  Natural key (e.g. customer_id) vs Hashed key (e.g. hash of order_id)

• Databases (SQL vs NoSQL)

  SQL (relational) vs NoSQL (document / key-value)

Networking & Delivery

• API Gateway

  Duplicated per service vs Centralized in a gateway

• CDN

  Static assets (images, CSS, JS) vs Personalized / dynamic responses

• Load Balancers

  Round robin vs Least-connections

• Message Queues

  Synchronous call vs Queued (async)

• Rate Limiting

  Fixed window vs Token bucket

Architecture Patterns

• Circuit Breaker

  No circuit breaker vs Circuit breaker

• CQRS

  One shared model vs CQRS (separate read/write models)

• Event-Driven Architecture

  Direct service calls vs Events (pub/sub)

• Microservices Architecture

  Monolith vs Microservices

• Monolith vs Microservices

  Splitting before the org is ready vs Splitting when team size forces it

Security

• Authentication vs Authorization

  Route-level authorization only vs Object-level authorization

• Encryption & TLS

  Encryption in transit (TLS) vs Encryption at rest

• JWT & Sessions

  Session (server-side lookup) vs JWT (self-contained, verified)

• OAuth 2.0 & OpenID Connect

  OAuth access token vs OIDC ID token

• Password Hashing

  Fast hash (SHA-256, MD5) vs Slow, salted hash (bcrypt, argon2)

Reliability

• Idempotency Keys

  With idempotency keys vs Without them

• Outbox Pattern

  Direct publish (dual write) vs Outbox pattern

Observability & SRE

• Alerting & On-Call

  Cause-based alerts vs Symptom-based alerts

• Chaos Engineering

  Never tested in production vs Tested via a chaos experiment

• Disaster Recovery

  Backups only vs Continuous replication + warm standby

• Distributed Tracing

  Manual propagation vs Shared middleware / instrumentation library

• Health Checks & Readiness Probes

  One combined health check vs Separate liveness and readiness checks

• Metrics, Logs & Traces

  Metrics + logs, uncorrelated vs Metrics + logs + traces, correlated by request

• SLIs, SLOs & Error Budgets

  No SLO, reliability by vibes vs SLO with an error budget

Web-scale

• Design a URL Shortener

  Base62 counter vs Hash + collision check

Java & Spring

Coming soon.

Payments

• Design a Payment Gateway with Idempotent Retries

  Postgres unique constraint vs Redis SETNX + TTL