jongminchungtech
KO

Series

Failure Handling in Distributed Systems

Design principles for ambiguous outcomes, Saga maintenance costs, and the reconciliation loops used by OpenStack and Kubernetes.

1. Failure Handling in Distributed SystemsexperimentalDistributed Failure Handling Part 1: How Far Did the Request Get?Analyze ambiguous outcomes caused by process termination and response loss, then establish a minimum safety boundary with transactions, idempotency, outbox, and reconciliation.2. Failure Handling in Distributed SystemsexperimentalDistributed Failure Handling Part 2: The Hidden Cost of Saga and OrchestrationAnalyze the transition, compensation, versioning, and operational costs that Saga, state machines, choreography, and orchestrators add outside the happy path.3. Failure Handling in Distributed SystemsexperimentalDistributed Failure Handling Part 3: How OpenStack and Kubernetes Converge After FailureCompare how OpenStack Nova and Kubernetes use durable state, asynchronous commands, idempotent reconciliation, and fencing instead of a global transaction.