Jordi Villar

Systems

Study notes on distributed systems and ClickHouse internals. Each one is a single page of short sections with simulators you can step through and break. They read in order, but each stands on its own.

Labs

  1. 00Commit and ConsensusTwo-phase and three-phase commit, Paxos and Raft, and how ZooKeeper and ClickHouse Keeper put them to work.Consensus
  2. 01Ordering, IDs and ClocksPhysical and logical clocks, TrueTime, ID generators, and how ClickHouse orders its own data with Keeper.Ordering
  3. 02Distributed FundamentalsWhat an acknowledgement promises: durability, idempotency, delivery semantics, failure detection and backpressure.Durability
  4. 03Workload SchedulingHow one ClickHouse server splits CPU, memory, IO and query slots between tenants, and the fairness math underneath.Scheduling
  5. 04Consistency Models and QuorumsWhat each guarantee allows, why R + W > N is weaker than it sounds, and what a ClickHouse SELECT actually sees.Consistency
  6. 05Storage and Fault ToleranceImmutable parts, merges, replication through Keeper, object storage, and recovering from lost or broken replicas.Storage
  7. 06Cloud OLAP DesignA worked design of a multi-tenant OLAP service on shared object storage, from SLOs to failure modes and cost.System design
  8. 07Concurrency and Production DebuggingFrom the C++ memory model and lock contention to incident drills that make you pick the next query to run.Concurrency
  9. 08Consensus GapsThe follow-ups after 2PC, 3PC and Raft: who blocks, what breaks under partitions, FLP, Paxos Commit and Raft reads.Consensus
  10. 09The System Design InterviewThe format and rubric, a seven-step method, pacing with a practice clock, estimation, deep dives, failure walks and a prompt bank.Interview