Tag
idempotency
- Jul 13, 2026 Anatomy of a Runbook Step Why a runbook step needs a precondition, a bounded action, and a verification that proves the system changed, and how steps fail when any of the three is missing. devops
- Jun 3, 2026 Exactly-Once Delivery Across System Boundaries The dual-write problem, the outbox that removes it, and the narrow sense in which a message can be applied exactly once when the database and the broker are separate systems. backend systems
- May 26, 2026 Reconciliation Loops: Why Level-Triggered Control Beats Event Handling A platform that compares desired state to observed state on every pass recovers from missed events and partial failures. The mechanism, its cost in API traffic, and the edge cases where a loop oscillates. virtualization
- Mar 13, 2026 Idempotency at the Effect Boundary, Not the Handler Entry Deduplicating a job at the point where it changes the world: unique constraints, request keys, delta-to-fact conversion, and the two-phase effect that survives a crash mid-call. backend systems
- Feb 6, 2026 Delivery Semantics and Where Duplicates Come From At-most-once, at-least-once, and exactly-once describe the disposition of a single message when an acknowledgement is lost, not a property of the network. backend systems
- Dec 30, 2025 Idempotent Task Boundaries and Atomic Output Publication How a task makes a second execution safe, why writing to a staging location and swapping is cheaper than deduplication, and what the swap cannot protect. data engineering
- Nov 24, 2025 Fan-Out Topologies and Duplicate Export What happens when one agent serves several backends, why a collector behind a load balancer exports each span twice, and how to keep retries from multiplying records. observability
- Oct 20, 2025 Serverless Execution and the Per-Event Invocation Model When the unit of deployment is a single event handler, lifecycle, concurrency, and retries stop being operational details and become the architecture. What the invocation model gives and what it takes away. software architecture
- Sep 4, 2025 Event Notification and Event-Carried State Transfer Compared Two events can carry the same name and different obligations: one says a fact changed and leaves the state behind, the other carries the state. The choice sets coupling, ordering needs, and deduplication cost. software architecture
- Aug 23, 2025 Domain Events: Publishing State Transitions Without Losing Them Why the database commit and the event publish are two writes that can disagree, how an outbox closes the gap, and what ordering and duplication guarantees consumers actually get. software architecture
- Jun 17, 2025 Specifying Partial Failure, Atomicity, and Retry Semantics How a contract states what survives a crash mid-operation, why idempotency keys must be chosen by the caller, and where at-least-once delivery forces deduplication. software engineering
- Jun 9, 2025 Durable Deferred Work and the Persistence Boundary How a work item survives process death, what must be written before an acknowledgement is sent, and why the durable record is the only scheduler that cannot be lost. mobile computing
- Apr 22, 2025 Idempotency Keys and the Unknown-Outcome Problem How a client-supplied key turns a retried mutation into at-most-once execution, where the key has to be stored, and which failure windows remain open. backend systems
- Apr 9, 2025 Cancellation Propagation and the Work That Outlives Its Caller Cancellation is a cooperative signal that reaches only the code that checks it, so a cancelled request can still commit a write; scoping the non-cancellable part is the design work. backend systems
- Mar 19, 2025 Resumable Backfills: Watermarks, Chunk Boundaries, and Gap-Free Progress How a long-running backfill records where it stopped, why the watermark must advance over a total order rather than a clock, and what idempotent chunks cost. distributed systems
- Jan 10, 2025 Where the Checkpoint Goes: Offset Ordering and Two Failure Directions Committing a read position before or after the sink write chooses between losing data and repeating it, and the choice propagates through every downstream consumer. data engineering
- Dec 7, 2024 Idempotency Keys and Duplicate Suppression at Service Boundaries How to make a retried request safe when the first attempt had an unknown outcome, where the claim record must sit relative to the side effect, and when key-based deduplication is not enough. software architecture
- Apr 5, 2024 Backfilling a New Store While the Old One Still Takes Writes How to copy historical rows into a replacement without a consistent cut, how keyset pagination and version guards handle concurrent updates, and what a reconciliation pass can and cannot prove. software architecture
- Jan 23, 2024 Method Semantics: Safe, Idempotent, and Cacheable Requests What the three method properties actually license a client, cache, or proxy to do, and where the definitions stop matching what real methods do. internet protocols
- Oct 28, 2023 Retry Classification When the Failure Arrives as a Timeout A timeout is the absence of an answer, not a statement that the request failed, and the decision to retry depends on whether the server can tell a duplicate from a first attempt. mobile computing
- Oct 18, 2023 Quarantine and Replay: What Happens to Records That Fail Validation Rejecting a record is a write to a second table, and replaying it later is a second write path; both need a schema, an ordering rule, and a bound on how long they stay open. data engineering
- Jul 28, 2023 CQRS and the Consistency Window Between Writes and Read Models Separating the write model from query models buys shape and contention relief, and pays for it with a lag the user can observe. Checkpoints, version tokens, and read-your-writes routing decide whether that lag is tolerable. software architecture
- Jun 26, 2023 What Exactly-Once Delivery Requires at Each Hop The guarantee is a conjunction of idempotence keys, transactional positions, and source cooperation, and it terminates at the first hop that lacks any of them. data engineering
- Mar 16, 2023 Idempotency Keys and the Retry Window for Writes How a client-supplied key turns a retried POST into a single effect, what the server has to store to honour it, and where the guarantee quietly expires. software architecture
- Jan 23, 2023 Retransmission and the Duplicate Request Cache A lost reply is indistinguishable from a lost request, so a client resends and a server must decide whether the operation already ran. The reply cache is the mechanism that turns at-least-once delivery into at-most-once execution. filesystems
- Dec 5, 2022 Row Identity and Why Deduplication Needs a Key, Not a Comparison Deciding that two records describe the same entity requires a declared key; comparing whole rows fails on late-arriving corrections, and windowed dedup only moves the ambiguity. data engineering