System design interview steps & template

Trade-offs:

  • CAP trade-off, availability vs consistency, Strong vs eventual consistency

    • Distributed database design, key-value design

  • Latency vs Throughput

  • ACID vs based database properties

  • Read-through and write-through case

  • Discovery service with gossip or centeraized metadata (bookkeeper)

Dive deep areas:

  • Potential bottlenecks of the system, like hot partition, scale

  • Potential failure cases: single node failure, network partition, split brain

  • Critical algorithms: consistent hashing, erasure coding, bloom filtering, geohashing, timeserial database

Common components for distributed basic services, like distributed key-value, database, message queue, streaming:

  • Load Balancer

  • API services / Control-plane: API service + authentication + authorization + metering + coordination with metadata service and placement service

  • Metadata service: metadata management, leader election, metadata caching

  • Placement service: decide the partition and entity placement logic

  • Workload management service: traffic movement, re-sharding, drain, block

  • Data-plane: cluster and nodes management, storage nodes + data caching, data migration, failure handling, data integrity,

Common components for distributed micro application service systems, like facebook, youtube, twiter, ticket systems

  • Load balancer

  • API services / Control-plane

  • Message Queue / Kafka Streaming

  • Distributed caching / local caching

  • Index builder to support better searching

  • Data Aggregator

  • CDN

Low level storage systems, like rocksDB, BDB, kafka file log:

https://artem.krylysov.com/blog/2023/04/19/how-rocksdb-works/

PreviousTopics 23NextTypical systems and tips

Last updated