query-plan-reader

Execution plan analysis reads operator costs, row estimate mismatches, and spills to identify the dominant bottleneck. It translates scan, join, and sort nodes into plain language, separating real performance issues from misleading assumptions.

Query tuning ranks fixes like statistics refresh, query rewrite, index changes, and schema adjustments by their estimated cost reduction and likely performance impact. This evidence-based approach prioritizes the cheapest effective solution.

Multi-tenant plan reading accounts for tenant predicates and row-security filters that influence index choices and operator costs. It evaluates whether the execution plan properly utilizes tenant-aware indexes under representative data volumes.

Estimate-versus-actual divergence occurs when optimizer statistics are outdated or parameter values skew row expectations. Analyzing this mismatch reveals whether the chosen scan or join operator is inefficient for the actual data scale.

Re-verification runs the query execution plan at the same parameter values and data scale after applying the fix. This confirms whether the new index or query rewrite actually changed the operator behavior and reduced cost.

Statistics refresh is needed when estimate-versus-actual rows diverge significantly without missing indexes. Index changes are justified when the execution plan shows expensive sequential scans or inefficient joins that a targeted index could eliminate.