Cardinality Estimation Error
The optimizer predicts row counts per node from per-column statistics, assuming independence between predicates and uniformity within histogram buckets; correlated columns, skewed values, and expressions on columns break those assumptions. A misestimate flips join algorithms and join order — the usual fixes are fresher or extended (multi-column) statistics, or rewriting the predicate so it is estimable.
This Concept is waiting for its first lesson!
The optimizer predicts row counts per node from per-column statistics, assuming independence between predicates and uniformity within histogram buckets; correlated columns, skewed values, and expressions on columns break those assumptions. A misestimate flips join algorithms and join order — the usual fixes are fresher or extended (multi-column) statistics, or rewriting the predicate so it is estimable.
Are you a teacher? Sign in to start contributing.
Sign In