What problem does it solve?
This Skill solves the problem of bulk-loading large datasets from object storage or HDFS into StarRocks with repeatable, production-friendly control over formats, authentication, and load monitoring.
Core Features & Use Cases
- Bulk Broker Load into StarRocks: Perform
LOAD LABEL from S3/HDFS/GCS/Azure/MinIO sources for large batch ETL without tying up the caller synchronously.
- Multi-format ingestion: Load CSV, Parquet, and ORC with schema mapping expressions and format-specific configuration.
- Operational safety & observability: Track progress via
SHOW LOAD, cancel jobs with CANCEL LOAD, and tune parallelism and timeouts for predictable throughput.
- Use Case: Daily/hourly landing-zone batch ETL where partitioned lake files (e.g.,
dt=YYYY-MM-DD/*.parquet) must be ingested into one or multiple StarRocks tables and then followed by ANALYZE for fresh optimizer stats.
Quick Start
Use the starrocks-broker-load Skill to load Parquet partition files from S3 into a target table by providing the correct LOAD LABEL statement, broker authentication, and PROPERTIES such as timeout for an async batch run.