AWS aurora-dsql documentation change
Summary
Removed preview status for CDC streams, added details about write-set compaction and net change capture, updated delivery semantics
Security assessment
Changes focus on feature maturity and technical implementation details. No security vulnerabilities or incidents are mentioned. Audit trail improvements are operational enhancements without explicit security context.
Diff
diff --git a/aurora-dsql/latest/userguide/cdc-streams.md b/aurora-dsql/latest/userguide/cdc-streams.md index f22ad320c..b7e4b828c 100644 --- a//aurora-dsql/latest/userguide/cdc-streams.md +++ b//aurora-dsql/latest/userguide/cdc-streams.md @@ -9 +9 @@ How it worksTopics on this pageRelated topicsOrdering and delivery semanticsMult -# Change data capture streams (Preview) +# Change data capture streams @@ -11,5 +11 @@ How it worksTopics on this pageRelated topicsOrdering and delivery semanticsMult -###### Important - -This feature is provided as an AWS Preview and is subject to change. For more information, see section 2, Betas and Previews, in the [AWS Service Terms](https://aws.amazon.com/service-terms/). To learn more about pricing for CDC streams, see the [Aurora DSQL pricing page](https://aws.amazon.com/rds/aurora/dsql/pricing/). - -Before general availability, we will add new operation types (`"op": "u"` for updates) to your stream payload. To ensure your application handles these changes without modification, treat any unrecognized `op` value as an upsert by applying the `after` payload. See [Understanding CDC records](./cdc-record-format.html) for details. +For pricing information, see the [Aurora DSQL pricing page](https://aws.amazon.com/rds/aurora/dsql/pricing/). @@ -17 +13 @@ Before general availability, we will add new operation types (`"op": "u"` for up -Amazon Aurora DSQL change data capture (CDC) streams committed database changes in near real time directly to Amazon Kinesis Data Streams. Aurora DSQL delivers each committed row-level change as a structured JSON record to a Kinesis data stream that you configure. +Amazon Aurora DSQL change data capture (CDC) streams committed database changes in near real time directly to Amazon Kinesis Data Streams. Aurora DSQL delivers the final committed state of each changed row as a structured JSON record to a Kinesis data stream that you configure. @@ -25 +21 @@ CDC is useful when you want to: - * **Maintain an audit trail** – Capture every committed change for compliance, debugging, or historical analysis. + * **Maintain an audit trail** – Capture the committed net effect of every transaction for compliance, debugging, or historical analysis. @@ -34 +30,3 @@ CDC is useful when you want to: -Aurora DSQL reads committed transactions, formats each row change as a structured JSON record, and delivers it to a Kinesis data stream that you configure. CDC automatically captures every `INSERT`, `UPDATE`, and `DELETE` across all user tables in the cluster. Apply filtering logic in your downstream apps by using the `source.schema` and `source.table` fields in each CDC record to focus on the tables or changes your app needs. +Aurora DSQL reads committed transactions, formats the net row-level change as a structured JSON record, and delivers it to a Kinesis data stream that you configure. CDC automatically captures the committed effect of every `INSERT`, `UPDATE`, and `DELETE` across all user tables in the cluster. Apply filtering logic in your downstream apps by using the `source.schema` and `source.table` fields in each CDC record to focus on the tables or changes your app needs. + +Aurora DSQL compacts the write set of each transaction before publishing CDC records and emits a single record for each row with a net change, reflecting the final state. For details, see [Write-set compaction](./cdc-record-format.html#cdc-write-set-compaction). @@ -68 +66,3 @@ CDC streams use a bring-your-own-target model. You create and manage the Kinesis -Aurora DSQL CDC guarantees that every committed change reaches the target at least one time. Aurora DSQL can deliver a record more than one time. Design your app to handle duplicates. You can identify a duplicate by comparing `source.ts_ns` and the primary key values—a duplicate has the same values as the original delivery. +Aurora DSQL produces one CDC record for each row with a net change at commit time. For details about how Aurora DSQL compacts multiple operations on the same row, see [Write-set compaction](./cdc-record-format.html#cdc-write-set-compaction). + +Aurora DSQL guarantees at-least-once delivery for every record it produces. Aurora DSQL can deliver the same record more than one time, so design your app to handle duplicates. You can identify a duplicate by comparing `source.ts_ns` and the primary key values—a duplicate has the same values as the original delivery. @@ -99 +99 @@ Track the highest `source.ts_ns` value per primary key. Discard any record with -Remove duplicates by comparing `source.ts_ns` combined with the primary key values. Buffer incoming records and sort by `source.ts_ns` before processing to reconstruct commit order. +Remove duplicates by comparing `source.ts_ns` combined with the primary key values. Buffer incoming records and sort by `source.ts_ns` before processing to reconstruct commit order. CDC delivers only the final state of each row per committed transaction (after [Write-set compaction](./cdc-record-format.html#cdc-write-set-compaction)). If your audit workflow requires a record for every SQL statement, perform each statement in its own transaction.