database/mariadb
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2025-07-29 05:21:33 +03:00
Code Activity

MDEV-16188 Introduced the notion of adjusted filter gain.

Browse Source 
Due to inconsistent usage of different cost models to calculate
the cost of ref accesses we have to make the calculation of the
gain promising by usage a range filter more complex.
This commit is contained in:
Igor Babaev
2019-02-14 00:15:48 -08:00
parent 76c34a74a8
commit e1de23b8d5
15 changed files with 181 additions and 92 deletions
Download Patch File Download Diff File Expand all files Collapse all files

63
sql/sql_select.cc
View File

@ -6981,6 +6981,7 @@ best_access_path(JOIN *join,
KEYUSE *hj_start_key= 0;
SplM_plan_info *spl_plan= 0;
Range_rowid_filter_cost_info *filter= 0;
double filter_cmp_gain= 0;
disable_jbuf= disable_jbuf || idx == join->const_tables;
@ -7376,12 +7377,14 @@ best_access_path(JOIN *join,
if (records < DBL_MAX)
{
double rows= record_count * records;
double access_cost_factor= MY_MIN(tmp / rows, 1.0);
filter=
table->best_range_rowid_filter_for_partial_join(start_key->key, rows);
table->best_range_rowid_filter_for_partial_join(start_key->key, rows,
access_cost_factor);
if (filter)
{
tmp-= filter->get_adjusted_gain(rows, s->worst_seeks) -
filter->get_cmp_gain(rows);
filter->get_cmp_gain(rows);
tmp-= filter->get_adjusted_gain(rows) - filter->get_cmp_gain(rows);
DBUG_ASSERT(tmp >= 0);
}
}
@ -7508,12 +7511,14 @@ best_access_path(JOIN *join,
if ( s->quick->get_type() == QUICK_SELECT_I::QS_TYPE_RANGE)
{
double rows= record_count * s->found_records;
double access_cost_factor= MY_MIN(tmp / rows, 1.0);
uint key_no= s->quick->index;
filter= s->table->best_range_rowid_filter_for_partial_join(key_no,
rows);
filter=
s->table->best_range_rowid_filter_for_partial_join(key_no, rows,
access_cost_factor);
if (filter)
{
tmp-= filter->get_gain(rows);
tmp-= filter->get_adjusted_gain(rows);
DBUG_ASSERT(tmp >= 0);
}
}
@ -7556,7 +7561,6 @@ best_access_path(JOIN *join,
}
}
double best_records= rnd_records;
/* Splitting technique cannot be used with join cache */
if (s->table->is_splittable())
tmp+= s->table->get_materialization_cost();
@ -7569,28 +7573,32 @@ best_access_path(JOIN *join,
tmp give us total cost of using TABLE SCAN
*/
double filter_cmp_gain= 0;
if (filter)
double best_filter_cmp_gain= 0;
if (best_filter)
{
filter_cmp_gain= filter->get_cmp_gain(record_count * s->found_records);
best_filter_cmp_gain= best_filter->get_cmp_gain(record_count * records);
}
if (best == DBL_MAX ||
(tmp + record_count/(double) TIME_FOR_COMPARE*rnd_records <
(best_key->is_for_hash_join() ? best_time :
best + record_count/(double) TIME_FOR_COMPARE*records -
filter_cmp_gain)))
best_filter_cmp_gain)))
{
/*
If the table has a range (s->quick is set) make_join_select()
will ensure that this will be used
*/
best= tmp;
records= best_records;
records= rnd_records;
best_key= 0;
best_filter= 0;
if (s->quick && s->quick->get_type() == QUICK_SELECT_I::QS_TYPE_RANGE)
if (s->quick && s->quick->get_type() == QUICK_SELECT_I::QS_TYPE_RANGE &&
filter)
{
best_filter= filter;
best_filter_cmp_gain= filter_cmp_gain;
}
/* range/index_merge/ALL/index access method are "independent", so: */
best_ref_depends_map= 0;
best_uses_jbuf= MY_TEST(!disable_jbuf && !((s->table->map &
@ -8091,14 +8099,22 @@ optimize_straight_join(JOIN *join, table_map join_tables)
for (JOIN_TAB **pos= join->best_ref + idx ; (s= *pos) ; pos++)
{
/* Find the best access method from 's' to the current partial plan */
POSITION *position= join->positions + idx;
/* Find the best access method from 's' to the current partial plan */
best_access_path(join, s, join_tables, idx, disable_jbuf, record_count,
join->positions + idx, &loose_scan_pos);
position, &loose_scan_pos);
/* compute the cost of the new plan extended with 's' */
record_count*= join->positions[idx].records_read;
read_time+= join->positions[idx].read_time +
record_count / (double) TIME_FOR_COMPARE;
record_count*= position->records_read;
double filter_cmp_gain= 0;
if (position->range_rowid_filter_info)
{
filter_cmp_gain=
position->range_rowid_filter_info->get_cmp_gain(record_count);
}
read_time+= position->read_time +
record_count / (double) TIME_FOR_COMPARE -
filter_cmp_gain;
advance_sj_state(join, join_tables, idx, &record_count, &read_time,
&loose_scan_pos);
@ -8107,7 +8123,7 @@ optimize_straight_join(JOIN *join, table_map join_tables)
if (use_cond_selectivity > 1)
pushdown_cond_selectivity= table_cond_selectivity(join, idx, s,
join_tables);
join->positions[idx].cond_selectivity= pushdown_cond_selectivity;
position->cond_selectivity= pushdown_cond_selectivity;
++idx;
}
@ -9006,8 +9022,15 @@ best_extension_by_limited_search(JOIN *join,
current_record_count= record_count * position->records_read;
else
current_record_count= DBL_MAX;
double filter_cmp_gain= 0;
if (position->range_rowid_filter_info)
{
filter_cmp_gain=
position->range_rowid_filter_info->get_cmp_gain(current_record_count);
}
current_read_time=read_time + position->read_time +
current_record_count / (double) TIME_FOR_COMPARE;
current_record_count / (double) TIME_FOR_COMPARE -
filter_cmp_gain;
advance_sj_state(join, remaining_tables, idx, &current_record_count,
&current_read_time, &loose_scan_pos);