OffNet Newsroom

Daily topic roundup

Database Technology

Saturday, July 11, 2026 · 7 stories, curated & summarized — click any story for the source.

Percona has released an experimental MySQL 9.7 build that allows tables to use DuckDB as a storage engine via ENGINE=DuckDB. This setup enables analytical queries to run directly on the source data within the same server process, eliminating the need for a separate copy. Benchmarks on TPC-H at scale factor 10 showed that while InnoDB timed out on six queries and consumed significant CPU, the DuckDB engine handled the workload differently.

  • Experimental MySQL 9.7 build supports ENGINE=DuckDB for analytical workloads.
  • Queries run in-process on the same server, avoiding data duplication.
  • TPC-H SF10 tests show InnoDB struggling with timeouts and CPU usage.
  • Designed for OLAP tasks within a standard MySQL connection context.
TRADE-OFFInnoDB vs DuckDB on TPC-HInnoDB StrugglesTimed out on six queriesConsumed significant CPU resourcesDuckDB EngineHandled workload differentlyRan analytical queries in-processvs
Planet PostgreSQL database

PostgreSQL enable_seqscan Does Not Disable Sequential Scans

The enable_seqscan GUC does not strictly prevent sequential scans; it merely lowers their cost weight to discourage the planner from choosing them. Developers often mistakenly believe setting this to off will force index usage, but the planner may still select a sequential scan if it is the only viable execution path. This behavior is by design, as full table scans are sometimes the only way to satisfy a query.

  • enable_seqscan only influences planner cost estimates, it does not enforce index-only scans
  • Setting off will not prevent sequential scans when no other execution plan exists
  • Relying on this GUC to force index usage is a common misconception that leads to performance issues
  • Use EXPLAIN ANALYZE to verify actual execution plans rather than assuming GUCs dictate behavior
AWS What's New awsdatabase

AWS DMS Schema Conversion adds offline SQL Server support

AWS Database Migration Service now allows offline schema conversion for Microsoft SQL Server, removing the need for direct connectivity to source databases. Users extract metadata using standard commands locally and upload it to DMS for processing, achieving the same results as the connected approach. This capability bypasses security reviews, firewall changes, and VPN setups that typically delay migration projects.

  • Eliminates need for direct DB connectivity, bypassing firewall and VPN setup delays.
  • Ideal for strict security policies restricting external tool access to production SQL Server.
  • Delivers identical conversion results to the online connected approach.
  • Allows internal security teams to review human-readable metadata files before upload.
HOW IT WORKSOffline SQL Server Migration Steps1Extract metadata locally2Review files internally3Upload to DMS4Convert schema offline
Planet PostgreSQL database

PostgreSQL 14 Throwing Errors Impossible for Its Version

A PostgreSQL 14 instance is generating an error message that the software version cannot logically produce. This indicates a mismatch between the binary code and the error catalog or internal state. The anomaly suggests potential corruption or a build artifact issue rather than standard query behavior.

  • Verify the exact binary build and patch level against the error definition.
  • Check for corrupted system catalogs or shared memory state.
  • Reproduce in a non-production environment to isolate the trigger.
  • Consider binary replacement if build artifacts are suspect.

An examination of governance challenges in financial services reveals that while individual teams like data engineering and AI development make sound, responsible decisions in isolation, significant risks emerge when viewing their interactions holistically. The C.A.L.M. platform test is introduced to identify these cross-functional misalignments, specifically where data movement for AI experimentation creates visibility gaps between engineering, AI, and compliance groups. This approach highlights how decentralized mandates can inadvertently lead to governance blind spots despite local adherence to best practices.

  • Siloed team decisions often mask cross-functional governance risks in AI data pipelines.
  • The C.A.L.M. framework exposes misalignments between data engineering and AI teams.
  • Compliance visibility gaps arise when data moves to isolated experimental environments.
  • Holistic reviews are necessary to catch issues invisible in individual team conversations.
AWS What's New awsdatabase ↺ since 07-09

Amazon Aurora DSQL CDC Generally Available for Real-Time Event Streaming

AWS has made change data capture for Amazon Aurora DSQL generally available, allowing real-time streaming of insert, update, and delete operations to Amazon Kinesis Data Streams. This managed feature captures database changes without requiring infrastructure management and is designed to have zero impact on database workload performance. Users can integrate these events into microservices, trigger Lambda functions, or deliver data to S3, Redshift, and OpenSearch via Firehose.

  • Stream real-time DSQL changes to Kinesis for event-driven architectures without managing infrastructure.
  • CDC captures inserts, updates, and deletes with zero impact on database performance.
  • Integrate with Lambda, S3, Redshift, and OpenSearch via Amazon Data Firehose.
  • Available in all regions supporting Aurora DSQL; no additional setup required.
HOW IT WORKSDSQL Change Data Flow1Aurora DSQL captures changes2Stream to Kinesis Data Streams3Trigger Lambda functions4Deliver to S3 or Redshift
AWS Database Blog awsdatabase ↺ since 07-08

RDS PostgreSQL 18 Boosts Logical Replication with Auto-Cleanup and Conflict Stats

Amazon RDS for PostgreSQL 18 introduces key logical replication enhancements, including default parallel streaming and automatic replication slot cleanup via idle_replication_slot_timeout. Users can now publish stored generated columns, toggle two-phase commit on active subscriptions, and monitor conflicts using new pg_stat_subscription_stats counters. These features are available for RDS for PostgreSQL 18.0+ and Aurora PostgreSQL.

  • Parallel streaming is enabled by default, likely improving replication throughput for large datasets.
  • Use idle_replication_slot_timeout to automatically clean up unused replication slots and prevent disk exhaustion.
  • Monitor subscription health with new conflict counters in pg_stat_subscription_stats for faster troubleshooting.
  • Publish stored generated columns using the new publish_generated_columns parameter for complex schemas.
  • Toggle two-phase commit on running subscriptions without restarting them, reducing operational downtime.

Fleet impact: For RDS PostgreSQL and Aurora fleets, enable idle_replication_slot_timeout immediately to prevent storage bloat from orphaned slots. Review pg_stat_subscription_stats to detect replication lag or conflicts early, and verify parallel streaming is active to optimize WAN-bound replication performance.