Faster Postgres reader (2×+ throughput)
The Postgres reader sustains 2x or more CDC throughput through improved compression and the removal of several O(N) hot paths in the replication reader.
Read moreChangelog
- Performance
- Performance
MySQL parallel backfill (integer primary keys)
MySQL tables with integer primary keys can backfill in parallel, with the table chunked into primary-key ranges that load concurrently.
Read more - Performance
Snowflake Turbo Mode (Dynamic Warehouse Scaling)
Snowflake pipelines switch to a higher-capacity warehouse when ingestion lag and workload thresholds are exceeded, then switch back once backlog clears.
Read more - Performance
Snowflake Transient Staging Tables
Snowflake staging tables are created as transient tables at runtime and dropped after use, which removes time travel and failsafe storage costs.
Read more - Performance
Soft Partitioning
Soft Partitioning routes incoming rows into time-based partition tables while maintaining a unified view that queries across all partitions.
Read more - Performance
Sub-Second Pipeline Deployment
Pipeline deployments complete in under 0.5 seconds each, down from 3 to 5 seconds, keeping large Terraform rollouts within execution limits.
Read more - Performance
Parallel Segmented Backfills for Postgres
Parallel Segmented Backfills split Postgres tables into row segments by integer primary key and backfill them in parallel as an alternative to CTID scans.
Read more - Performance
Backfill Tuning: Picking the Right Batch Size
Backfill batch size is now configurable, and the default rises from 5,000 to 25,000 rows per chunk.
Read more