Transactions across machines

Reason about atomic commit, isolation, concurrency conflicts, and recovery across partitions without confusing availability with atomicity.

10 published lessons1 module260 min of guided study
  1. 01
    Transactions across machinesadvanced · 10 lessons

Follow the path or begin with a question. Every lesson is open.

Your notebook0 of 10 lessons completed

Your place, at your pace. Progress is optional and saved only in this browser when available.

Inside this module

Each lesson includes an interactive lab

Transactions across machines

Reason about atomic commit, isolation, concurrency conflicts, and recovery across partitions without confusing availability with atomicity.

  1. 01

    Atomic commit and two-phase commit

    Separate agreement on a cross-shard outcome from isolation, replication, and the availability of each participant.

    26 min
  2. 02

    Coordinator failure and decision recovery

    Use durable state and transaction identity to resolve prepared work without inventing outcomes after a timeout.

    26 min
  3. 03

    Detect deadlocks across shard boundaries

    Build a global wait-for argument while separating real cycles, stale observations, timeouts, and transaction generations.

    26 min
  4. 04

    Timestamp ordering and validation

    Use logical order and read-version evidence without mistaking clocks or a validation snapshot for a complete commit protocol.

    26 min
  5. 05

    Serializable snapshot isolation

    Read multiversion dependencies, recognize write skew, and understand SSI’s conservative detection of dangerous structures.

    26 min
  6. 06

    Deterministic transaction ordering

    Move agreement before execution while preserving access-set, replay, and nondeterminism assumptions.

    26 min
  7. 07

    Two-phase locking across shards

    Connect conflict serializability, lock lifetime, atomic commit, and predicate protection without confusing 2PL with 2PC.

    26 min
  8. 08

    Retries and ambiguous transaction outcomes

    Separate known aborts from lost commit replies, and design bounded logical retries with durable idempotency evidence.

    26 min
  9. 09

    Sagas and compensating actions

    Design recoverable workflows whose committed steps can interleave, without describing semantic compensation as isolated rollback.

    26 min
  10. 10

    Test cross-shard invariants and recovery histories

    Construct fault schedules and history oracles that distinguish atomicity, isolation, durability, and conditional progress.

    26 min
All engineering modules