Higher-throughput Snowpipe Streaming

Snowpipe Streaming can now fan rows across stable channels using the configured channel limit, drain those channels concurrently, and retain per-channel retry and commit behavior. Related Snowflake changes make pipe creation idempotent, keep retries deterministic, match pipe columns by name, and apply streaming to both history and replication modes.

Why this matters

• Increase Snowflake ingestion throughput for larger pipelines
• Preserve key ordering within a channel while using parallel writes
• Reduce duplicate or failed writes during retries and schema changes