System Architecture Exercises
System Architecture - Practice Exercises
Exercises focused on system-level topology choices (microservices, modular monolith, event-driven, micro-kernel).
---
Foundations
Q: Explain the difference between a modular monolith and microservices.
A: A modular monolith is a single deployable with strong internal boundaries, while microservices are independently deployed services with separate data ownership. The monolith is simpler operationally; microservices scale and deploy independently but introduce distributed complexity.
---
Scenario Decisions
Q: You are a 6-person team building a trading platform MVP. Which architecture should you start with and why?
A: Start with a modular monolith. It enables fast delivery, strong consistency, and simpler operations. Design module boundaries so you can split services later if needed.
---
Q: You need to process millions of price ticks per second and fan out to multiple consumers. Which system style fits best?
A: Event-driven architecture. Use a broker (Kafka/RabbitMQ) to publish ticks and let consumers scale independently.
---
Design Exercises
Q: Sketch a high-level system architecture for an order execution system using microservices.
A:
[API Gateway]
| | |
[Orders] [Risk] [Pricing]
| | |
[DB] [DB] [Cache]
|
[Broker] -> [Audit]---
Q: Explain how you would enforce data ownership in microservices.
A: Each service owns its database and schema. Other services access data via APIs or events, not direct DB access. Cross-service workflows use sagas and events.
---
Implementation Drill
Q: Describe how you would implement an outbox pattern to keep DB and events consistent.
A: Write domain changes and an outbox record in the same transaction. A background worker reads the outbox table and publishes events to the broker, marking them as sent.
---
Advanced Trade-offs
Q: What operational capabilities do you need before moving to microservices?
A: Observability (logging, metrics, tracing), mature CI/CD, automated testing, service discovery, resilient networking, and on-call maturity.
---
Q: Where does a micro-kernel architecture make sense in trading systems?
A: When you need a stable core with customizable plugins, such as client-specific risk engines or pricing strategies.
---
Trading Scenario
Q: You are asked to move a trading system to microservices but latency is critical. How do you respond?
A: Start with a modular monolith and prove performance. If microservices are required, keep latency-sensitive paths in fewer services, use async messaging for non-critical flows, and benchmark end-to-end latency.
---
Total Exercises: 10+