System design · Deep dives

The techniques that keep coming up

6 ideas interviewers circle back to. Watch each one play out, push it in a widget, then keep the trade-offs.

Got a minute?12 one-minute shorts on these techniques. One idea per card, swipe for the next.

Open shorts
  1. 01 20 min

    Caching Techniques

    A cache trades freshness and memory for speed. The technique you choose decides what happens when the cache is stale, cold, or down.

    cache-asidewrite-throughwrite-behindinvalidationOpen the deep dive
  2. 02 20 min

    Sharding Strategies

    Sharding splits data across machines so no single database limits storage or write throughput. The shard key you pick is the most important decision.

    shardingconsistent hashingrendezvoussnowflakeOpen the deep dive
  3. 03 20 min

    Storing Data: Choosing the Right Store

    No single database is best at everything. Match each kind of data to the store whose strengths fit its access pattern, and keep one source of truth per fact.

    postgresmongodbdynamodbcassandraOpen the deep dive
  4. 04 20 min

    Polling, SSE and WebSockets

    HTTP is request and response, but users expect live updates. Each technique trades simplicity against latency and server cost.

    pollinglong pollingssewebsocketsOpen the deep dive
  5. 05 20 min

    Connections: Pooling, Keep-Alive and Limits

    Connections are expensive and limited. Most production outages from load come from exhausted pools, missing timeouts, or unbounded concurrency.

    connection poolpgbouncerkeep-alivetimeoutsOpen the deep dive
  6. 06 20 min

    Request Flows End to End

    Interviewers want to hear the path a request takes. Know the read, write, async, and failure flows and what each layer contributes.

    read pathwrite pathoutboxidempotencyOpen the deep dive