Allow estimate_num_groups() to pass back further details about the estimation

Here we add a new output parameter to estimate_num_groups() to allow it to
inform the caller of additional, possibly useful information about the
estimation.

The new output parameter is a struct that currently contains just a single
field with a set of flags.  This was done rather than having the flags as
an output parameter to allow future fields to be added without having to
change the signature of the function at a later date when we want to pass
back further information that might not be suitable to store in the flags
field.

It seems reasonable that one day in the future that the planner would want
to know more about the estimation. For example, how many individual sets
of statistics was the estimation generated from?  The planner may want to
take that into account if we ever want to consider risks as well as costs
when generating plans.

For now, there's only 1 flag we set in the flags field.  This is to
indicate if the estimation fell back on using the hard-coded constants in
any part of the estimation. Callers may like to change their behavior if
this is set, and this gives them the ability to do so.  Callers may pass
the flag pointer as NULL if they have no interest in obtaining any
additional information about the estimate.

We're not adding any actual usages of these flags here.  Some follow-up
commits will make use of this feature.  Additionally, we're also not
making any changes to add support for clauselist_selectivity() and
clauselist_selectivity_ext().  However, if this is required in the future
then the same struct being added here should be fine to use as a new
output argument for those functions too.

Author: David Rowley
Discussion: https://postgr.es/m/CAApHDvqQqpk=1W-G_ds7A9CsXX3BggWj_7okinzkLVhDubQzjA@mail.gmail.com
This commit is contained in:
David Rowley 2021-03-30 20:52:46 +13:00
parent efd9d92bb3
commit ed934d4fa3
8 changed files with 49 additions and 8 deletions

View File

@ -3087,7 +3087,7 @@ estimate_path_cost_size(PlannerInfo *root,
numGroups = estimate_num_groups(root,
get_sortgrouplist_exprs(root->parse->groupClause,
fpinfo->grouped_tlist),
input_rows, NULL);
input_rows, NULL, NULL);
/*
* Get the retrieved_rows and rows estimates. If there are HAVING

View File

@ -1969,7 +1969,8 @@ cost_incremental_sort(Path *path,
/* Estimate number of groups with equal presorted keys. */
if (!unknown_varno)
input_groups = estimate_num_groups(root, presortedExprs, input_tuples, NULL);
input_groups = estimate_num_groups(root, presortedExprs, input_tuples,
NULL, NULL);
group_tuples = input_tuples / input_groups;
group_input_run_cost = input_run_cost / input_groups;

View File

@ -1990,6 +1990,7 @@ adjust_rowcount_for_semijoins(PlannerInfo *root,
nunique = estimate_num_groups(root,
sjinfo->semi_rhs_exprs,
nraw,
NULL,
NULL);
if (rowcount > nunique)
rowcount = nunique;

View File

@ -3702,7 +3702,8 @@ get_number_of_groups(PlannerInfo *root,
double numGroups = estimate_num_groups(root,
groupExprs,
path_rows,
&gset);
&gset,
NULL);
gs->numGroups = numGroups;
rollup->numGroups += numGroups;
@ -3727,7 +3728,8 @@ get_number_of_groups(PlannerInfo *root,
double numGroups = estimate_num_groups(root,
groupExprs,
path_rows,
&gset);
&gset,
NULL);
gs->numGroups = numGroups;
gd->dNumHashGroups += numGroups;
@ -3743,7 +3745,7 @@ get_number_of_groups(PlannerInfo *root,
target_list);
dNumGroups = estimate_num_groups(root, groupExprs, path_rows,
NULL);
NULL, NULL);
}
}
else if (parse->groupingSets)
@ -4792,7 +4794,7 @@ create_distinct_paths(PlannerInfo *root,
parse->targetList);
numDistinctRows = estimate_num_groups(root, distinctExprs,
cheapest_input_path->rows,
NULL);
NULL, NULL);
}
/*

View File

@ -338,6 +338,7 @@ recurse_set_operations(Node *setOp, PlannerInfo *root,
*pNumGroups = estimate_num_groups(subroot,
get_tlist_exprs(subquery->targetList, false),
subpath->rows,
NULL,
NULL);
}
}

View File

@ -1713,6 +1713,7 @@ create_unique_path(PlannerInfo *root, RelOptInfo *rel, Path *subpath,
pathnode->path.rows = estimate_num_groups(root,
sjinfo->semi_rhs_exprs,
rel->rows,
NULL,
NULL);
numCols = list_length(sjinfo->semi_rhs_exprs);

View File

@ -3241,6 +3241,7 @@ typedef struct
Node *var; /* might be an expression, not just a Var */
RelOptInfo *rel; /* relation it belongs to */
double ndistinct; /* # distinct values */
bool isdefault; /* true if DEFAULT_NUM_DISTINCT was used */
} GroupVarInfo;
static List *
@ -3287,6 +3288,7 @@ add_unique_group_var(PlannerInfo *root, List *varinfos,
varinfo->var = var;
varinfo->rel = vardata->rel;
varinfo->ndistinct = ndistinct;
varinfo->isdefault = isdefault;
varinfos = lappend(varinfos, varinfo);
return varinfos;
}
@ -3311,6 +3313,12 @@ add_unique_group_var(PlannerInfo *root, List *varinfos,
* pgset - NULL, or a List** pointing to a grouping set to filter the
* groupExprs against
*
* Outputs:
* estinfo - When passed as non-NULL, the function will set bits in the
* "flags" field in order to provide callers with additional information
* about the estimation. Currently, we only set the SELFLAG_USED_DEFAULT
* bit if we used any default values in the estimation.
*
* Given the lack of any cross-correlation statistics in the system, it's
* impossible to do anything really trustworthy with GROUP BY conditions
* involving multiple Vars. We should however avoid assuming the worst
@ -3358,7 +3366,7 @@ add_unique_group_var(PlannerInfo *root, List *varinfos,
*/
double
estimate_num_groups(PlannerInfo *root, List *groupExprs, double input_rows,
List **pgset)
List **pgset, EstimationInfo *estinfo)
{
List *varinfos = NIL;
double srf_multiplier = 1.0;
@ -3366,6 +3374,10 @@ estimate_num_groups(PlannerInfo *root, List *groupExprs, double input_rows,
ListCell *l;
int i;
/* Zero the estinfo output parameter, if non-NULL */
if (estinfo != NULL)
memset(estinfo, 0, sizeof(EstimationInfo));
/*
* We don't ever want to return an estimate of zero groups, as that tends
* to lead to division-by-zero and other unpleasantness. The input_rows
@ -3577,6 +3589,14 @@ estimate_num_groups(PlannerInfo *root, List *groupExprs, double input_rows,
if (relmaxndistinct < varinfo2->ndistinct)
relmaxndistinct = varinfo2->ndistinct;
relvarcount++;
/*
* When varinfo2's isdefault is set then we'd better set
* the SELFLAG_USED_DEFAULT bit in the EstimationInfo.
*/
if (estinfo != NULL && varinfo2->isdefault)
estinfo->flags |= SELFLAG_USED_DEFAULT;
}
/* we're done with this relation */

View File

@ -68,6 +68,20 @@
p = 1.0; \
} while (0)
/*
* A set of flags which some selectivity estimation functions can pass back to
* callers to provide further details about some assumptions which were made
* during the estimation.
*/
#define SELFLAG_USED_DEFAULT (1 << 0) /* Estimation fell back on one
* of the DEFAULTs as defined
* above. */
typedef struct EstimationInfo
{
uint32 flags; /* Flags, as defined above to mark special
* properties of the estimation. */
} EstimationInfo;
/* Return data from examine_variable and friends */
typedef struct VariableStatData
@ -197,7 +211,8 @@ extern void mergejoinscansel(PlannerInfo *root, Node *clause,
Selectivity *rightstart, Selectivity *rightend);
extern double estimate_num_groups(PlannerInfo *root, List *groupExprs,
double input_rows, List **pgset);
double input_rows, List **pgset,
EstimationInfo *estinfo);
extern void estimate_hash_bucket_stats(PlannerInfo *root,
Node *hashkey, double nbuckets,