Time and Ordering in Distributed Systems
Choose physical, logical, vector, and hybrid clocks by the ordering facts a distributed protocol actually needs.
2026-07-14 · 14 min read
Choose physical, logical, vector, and hybrid clocks by the ordering facts a distributed protocol actually needs.
2026-07-14 · 14 min read
Make lease-based coordination safe by fencing stale owners at every protected resource with monotonic tokens.
2026-06-29 · 14 min read
Design distributed request paths around tail latency with honest percentiles, explicit budgets, bounded hedging, and deadline-aware retries.
2026-06-25 · 13 min read
Derive convergent replicated data from merge laws, causal context, and explicit product semantics.
2026-06-17 · 14 min read
Choose queues, streams, and durable logs by their ownership, retention, replay, ordering, and consumer-state contracts.
2026-06-05 · 12 min read
Design service discovery as a converging membership system with explicit readiness, liveness, staleness, and observability contracts.
2026-05-24 · 14 min read
Safe retries with idempotency records, payload hashes, transactions, inbox/outbox patterns, and honest delivery guarantees.
2026-05-16 · 13 min read
Choose request-placement algorithms by workload cost, locality, health evidence, churn, and hotspot behavior rather than by name alone.
2026-05-12 · 15 min read
Compose time budgets, circuit breakers, bulkheads, retry budgets, and degradation to contain service-level failures.
2026-04-30 · 13 min read
Build scalable membership from epidemic dissemination, suspicion evidence, phi accrual, incarnation handling, and convergence tests.
2026-04-18 · 14 min read
Two-phase commit turns durable participant votes into one atomic outcome, at the cost of blocking, coordination, and careful recovery.
2026-04-06 · 14 min read
Separate coordination from request execution with versioned desired state, reconciliation, bounded stale operation, and distinct security boundaries.
2026-04-03 · 12 min read