starrocks-explain-plan

What problem does it solve?

It helps you pinpoint why a StarRocks query is slow by turning EXPLAIN output into an actionable checklist for shuffle cost, scan selectivity, runtime filters, and optimizer decisions.

Core Features & Use Cases

• EXPLAIN mode selection for the right layer: guides you through logical (EXPLAIN), physical (EXPLAIN VERBOSE), costed (EXPLAIN COSTS), and runtime-with-metrics (EXPLAIN ANALYZE) views.
• Plan structure decoding: explains fragments, pipelines, and ExchangeNode communication types to identify where distributed cost is introduced.
• Root-cause signals for performance tuning: interprets partition pruning, MV/rollup rewrites, PREAGGREGATION behavior, runtime filter propagation, join strategies (BROADCAST/SHUFFLE/COLOCATE), and cardinality/statistics issues.
• Operator-level bottleneck analysis: shows how to use metrics like OperatorTotalTime, PeakMemoryBytes, and row movement (Pull/Push) to locate the slowest operator.

Quick Start

Run EXPLAIN ANALYZE for your slow StarRocks query, then check partition pruning in every OlapScanNode, confirm runtime filter IDs match between HashJoinNode and scan nodes, and verify whether the scan is reading from an MV rollup via the rollup/table fields.