diff options
| -rw-r--r-- | doc/src/sgml/planstats.sgml | 12 |
1 files changed, 7 insertions, 5 deletions
diff --git a/doc/src/sgml/planstats.sgml b/doc/src/sgml/planstats.sgml index 43ad57253ef..c7ec749d0a6 100644 --- a/doc/src/sgml/planstats.sgml +++ b/doc/src/sgml/planstats.sgml @@ -389,18 +389,20 @@ tablename | null_frac | n_distinct | most_common_vals </programlisting> In this case there is no <acronym>MCV</acronym> information for - <structfield>unique2</structfield> because all the values appear to be - unique, so we use an algorithm that relies only on the number of - distinct values for both relations together with their null fractions: + <structname>unique2</structname> and all the values appear to be + unique (n_distinct = -1), so we use an algorithm that relies on the row + count estimates for both relations (num_rows, not shown, but "tenk") + together with the column null fractions (zero for both): <programlisting> -selectivity = (1 - null_frac1) * (1 - null_frac2) * min(1/num_distinct1, 1/num_distinct2) +selectivity = (1 - null_frac1) * (1 - null_frac2) / max(num_rows1, num_rows2) = (1 - 0) * (1 - 0) / max(10000, 10000) = 0.0001 </programlisting> This is, subtract the null fraction from one for each of the relations, - and divide by the maximum of the numbers of distinct values. + and divide by the row count of the larger relation (this value does get + scaled in the non-unique case). The number of rows that the join is likely to emit is calculated as the cardinality of the Cartesian product of the two inputs, multiplied by the |