↓ Ir para o conteúdo principal
Scholion
Scholion

← todas as notas

📎 Webclip

System Design — Tips

The post collects system design tips around core trade-offs: understand functional and non-functional requirements, pick an architecture, and then choose patterns for scale, caching, storage, availability, security, and communication. It also pairs each choice with a concrete scenario such as read-heavy, write-heavy, low-latency, or globally distributed systems.

Reading notes
#

  • Define both functional and non-functional requirements before choosing the design.
  • Choose between monolith and microservices, and consider layered architecture for separation of concerns.
  • Scale horizontally, vertically, and with load balancing.
  • For databases, balance normalization and denormalization, choose SQL or NoSQL based on the data, and use sharding when needed.
  • For distributed locks, set a TTL, renew locks when necessary, and use tools like Redis, ZooKeeper, or etcd.
  • Use caching and CDNs to reduce latency and database load.
  • Build for fault tolerance with redundancy, failover, backup, and recovery.
  • Secure the system with authentication, authorization, encryption, firewalls, and intrusion detection.
  • Monitor performance and keep logs for debugging and analysis.
  • Use RESTful APIs for stateless services and message queues for asynchronous communication.
  • For low-latency video delivery, combine cache and CDN, with multi-region CDN for scale.
  • For read-heavy systems, cache popular data with TTL.
  • For write-heavy systems, use message queues and background workers.
  • Use NoSQL for unstructured data, SQL for ACID transactions, and object storage for videos, images, and files.
  • Use a load balancer for high availability and an API gateway for microservices.
  • Add indexes for faster queries, checksum algorithms for integrity, data lakes or append-only databases for analytics and audit trails, and data replication for fault tolerance.
  • Use WebSockets for real-time chat, heartbeats for failure detection, consistent hashing for cluster scaling, gossip protocol for decentralized updates, eventual consistency for availability, pagination for large lists, and autoscaling for traffic spikes.