starrocks-cbo

What problem does it solve?

This skill addresses slow or inefficient StarRocks queries caused by stale or missing optimizer statistics, leading to wrong join strategies, bad cardinality estimates, and poor execution plans.

Core Features & Use Cases

• ANALYZE TABLE for CBO statistics: Collect full, sampled, histogram, predicate-column, and partition-level statistics to refresh the cost model after ETL or bulk loads.
• Auto analyze configuration & monitoring: Enable and tune automatic statistics collection, then verify outcomes using SHOW ANALYZE STATUS and targeted maintenance workflows.
• Diagnose plan issues with EXPLAIN COSTS: Identify cardinality estimation errors and apply fixes via updated stats, histograms, or controlled optimizer hints (LEADING, BROADCAST/SHUFFLE, SET_VAR).
• Multi-column statistics for correlated filters: Improve selectivity estimation when single-column stats fail due to column correlation (StarRocks 3.5+).

Use case example: after a daily bulk load into an orders table, activate this skill to re-ANALYZE the affected partitions (and add histograms for skewed columns) so that join order and join type chosen by the CBO match real data distribution.

Quick Start

Use the starrocks-cbo skill to run ANALYZE TABLE for the affected orders table (including histogram or sampled stats as appropriate) and then confirm the improvement with EXPLAIN COSTS on your problematic join query.