1
0
mirror of https://github.com/postgres/postgres.git synced 2025-11-01 21:31:19 +03:00

Faster partition pruning

Add a new module backend/partitioning/partprune.c, implementing a more
sophisticated algorithm for partition pruning.  The new module uses each
partition's "boundinfo" for pruning instead of constraint exclusion,
based on an idea proposed by Robert Haas of a "pruning program": a list
of steps generated from the query quals which are run iteratively to
obtain a list of partitions that must be scanned in order to satisfy
those quals.

At present, this targets planner-time partition pruning, but there exist
further patches to apply partition pruning at execution time as well.

This commit also moves some definitions from include/catalog/partition.h
to a new file include/partitioning/partbounds.h, in an attempt to
rationalize partitioning related code.

Authors: Amit Langote, David Rowley, Dilip Kumar
Reviewers: Robert Haas, Kyotaro Horiguchi, Ashutosh Bapat, Jesper Pedersen.
Discussion: https://postgr.es/m/098b9c71-1915-1a2a-8d52-1a7a50ce79e8@lab.ntt.co.jp
This commit is contained in:
Alvaro Herrera
2018-04-06 16:23:04 -03:00
parent 11523e860f
commit 9fdb675fc5
27 changed files with 3993 additions and 415 deletions

View File

@@ -2150,6 +2150,38 @@ _copyMergeAction(const MergeAction *from)
return newnode;
}
/*
* _copyPartitionPruneStepOp
*/
static PartitionPruneStepOp *
_copyPartitionPruneStepOp(const PartitionPruneStepOp *from)
{
PartitionPruneStepOp *newnode = makeNode(PartitionPruneStepOp);
COPY_SCALAR_FIELD(step.step_id);
COPY_SCALAR_FIELD(opstrategy);
COPY_NODE_FIELD(exprs);
COPY_NODE_FIELD(cmpfns);
COPY_BITMAPSET_FIELD(nullkeys);
return newnode;
}
/*
* _copyPartitionPruneStepCombine
*/
static PartitionPruneStepCombine *
_copyPartitionPruneStepCombine(const PartitionPruneStepCombine *from)
{
PartitionPruneStepCombine *newnode = makeNode(PartitionPruneStepCombine);
COPY_SCALAR_FIELD(step.step_id);
COPY_SCALAR_FIELD(combineOp);
COPY_NODE_FIELD(source_stepids);
return newnode;
}
/* ****************************************************************
* relation.h copy functions
*
@@ -2277,21 +2309,6 @@ _copyAppendRelInfo(const AppendRelInfo *from)
return newnode;
}
/*
* _copyPartitionedChildRelInfo
*/
static PartitionedChildRelInfo *
_copyPartitionedChildRelInfo(const PartitionedChildRelInfo *from)
{
PartitionedChildRelInfo *newnode = makeNode(PartitionedChildRelInfo);
COPY_SCALAR_FIELD(parent_relid);
COPY_NODE_FIELD(child_rels);
COPY_SCALAR_FIELD(part_cols_updated);
return newnode;
}
/*
* _copyPlaceHolderInfo
*/
@@ -5076,6 +5093,12 @@ copyObjectImpl(const void *from)
case T_MergeAction:
retval = _copyMergeAction(from);
break;
case T_PartitionPruneStepOp:
retval = _copyPartitionPruneStepOp(from);
break;
case T_PartitionPruneStepCombine:
retval = _copyPartitionPruneStepCombine(from);
break;
/*
* RELATION NODES
@@ -5095,9 +5118,6 @@ copyObjectImpl(const void *from)
case T_AppendRelInfo:
retval = _copyAppendRelInfo(from);
break;
case T_PartitionedChildRelInfo:
retval = _copyPartitionedChildRelInfo(from);
break;
case T_PlaceHolderInfo:
retval = _copyPlaceHolderInfo(from);
break;

View File

@@ -915,16 +915,6 @@ _equalAppendRelInfo(const AppendRelInfo *a, const AppendRelInfo *b)
return true;
}
static bool
_equalPartitionedChildRelInfo(const PartitionedChildRelInfo *a, const PartitionedChildRelInfo *b)
{
COMPARE_SCALAR_FIELD(parent_relid);
COMPARE_NODE_FIELD(child_rels);
COMPARE_SCALAR_FIELD(part_cols_updated);
return true;
}
static bool
_equalPlaceHolderInfo(const PlaceHolderInfo *a, const PlaceHolderInfo *b)
{
@@ -3230,9 +3220,6 @@ equal(const void *a, const void *b)
case T_AppendRelInfo:
retval = _equalAppendRelInfo(a, b);
break;
case T_PartitionedChildRelInfo:
retval = _equalPartitionedChildRelInfo(a, b);
break;
case T_PlaceHolderInfo:
retval = _equalPlaceHolderInfo(a, b);
break;

View File

@@ -2156,6 +2156,17 @@ expression_tree_walker(Node *node,
return true;
}
break;
case T_PartitionPruneStepOp:
{
PartitionPruneStepOp *opstep = (PartitionPruneStepOp *) node;
if (walker((Node *) opstep->exprs, context))
return true;
}
break;
case T_PartitionPruneStepCombine:
/* no expression subnodes */
break;
case T_JoinExpr:
{
JoinExpr *join = (JoinExpr *) node;
@@ -2958,6 +2969,20 @@ expression_tree_mutator(Node *node,
return (Node *) newnode;
}
break;
case T_PartitionPruneStepOp:
{
PartitionPruneStepOp *opstep = (PartitionPruneStepOp *) node;
PartitionPruneStepOp *newnode;
FLATCOPY(newnode, opstep, PartitionPruneStepOp);
MUTATE(newnode->exprs, opstep->exprs, List *);
return (Node *) newnode;
}
break;
case T_PartitionPruneStepCombine:
/* no expression sub-nodes */
return (Node *) copyObject(node);
case T_JoinExpr:
{
JoinExpr *join = (JoinExpr *) node;

View File

@@ -1710,6 +1710,28 @@ _outFromExpr(StringInfo str, const FromExpr *node)
WRITE_NODE_FIELD(quals);
}
static void
_outPartitionPruneStepOp(StringInfo str, const PartitionPruneStepOp *node)
{
WRITE_NODE_TYPE("PARTITIONPRUNESTEPOP");
WRITE_INT_FIELD(step.step_id);
WRITE_INT_FIELD(opstrategy);
WRITE_NODE_FIELD(exprs);
WRITE_NODE_FIELD(cmpfns);
WRITE_BITMAPSET_FIELD(nullkeys);
}
static void
_outPartitionPruneStepCombine(StringInfo str, const PartitionPruneStepCombine *node)
{
WRITE_NODE_TYPE("PARTITIONPRUNESTEPCOMBINE");
WRITE_INT_FIELD(step.step_id);
WRITE_ENUM_FIELD(combineOp, PartitionPruneCombineOp);
WRITE_NODE_FIELD(source_stepids);
}
static void
_outOnConflictExpr(StringInfo str, const OnConflictExpr *node)
{
@@ -2261,7 +2283,6 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
WRITE_NODE_FIELD(full_join_clauses);
WRITE_NODE_FIELD(join_info_list);
WRITE_NODE_FIELD(append_rel_list);
WRITE_NODE_FIELD(pcinfo_list);
WRITE_NODE_FIELD(rowMarks);
WRITE_NODE_FIELD(placeholder_list);
WRITE_NODE_FIELD(fkey_list);
@@ -2286,6 +2307,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
WRITE_INT_FIELD(wt_param_id);
WRITE_BITMAPSET_FIELD(curOuterRels);
WRITE_NODE_FIELD(curOuterParams);
WRITE_BOOL_FIELD(partColsUpdated);
}
static void
@@ -2335,6 +2357,7 @@ _outRelOptInfo(StringInfo str, const RelOptInfo *node)
WRITE_NODE_FIELD(joininfo);
WRITE_BOOL_FIELD(has_eclass_joins);
WRITE_BITMAPSET_FIELD(top_parent_relids);
WRITE_NODE_FIELD(partitioned_child_rels);
}
static void
@@ -2559,16 +2582,6 @@ _outAppendRelInfo(StringInfo str, const AppendRelInfo *node)
WRITE_OID_FIELD(parent_reloid);
}
static void
_outPartitionedChildRelInfo(StringInfo str, const PartitionedChildRelInfo *node)
{
WRITE_NODE_TYPE("PARTITIONEDCHILDRELINFO");
WRITE_UINT_FIELD(parent_relid);
WRITE_NODE_FIELD(child_rels);
WRITE_BOOL_FIELD(part_cols_updated);
}
static void
_outPlaceHolderInfo(StringInfo str, const PlaceHolderInfo *node)
{
@@ -3973,6 +3986,12 @@ outNode(StringInfo str, const void *obj)
case T_MergeAction:
_outMergeAction(str, obj);
break;
case T_PartitionPruneStepOp:
_outPartitionPruneStepOp(str, obj);
break;
case T_PartitionPruneStepCombine:
_outPartitionPruneStepCombine(str, obj);
break;
case T_Path:
_outPath(str, obj);
break;
@@ -4114,9 +4133,6 @@ outNode(StringInfo str, const void *obj)
case T_AppendRelInfo:
_outAppendRelInfo(str, obj);
break;
case T_PartitionedChildRelInfo:
_outPartitionedChildRelInfo(str, obj);
break;
case T_PlaceHolderInfo:
_outPlaceHolderInfo(str, obj);
break;

View File

@@ -1331,6 +1331,32 @@ _readOnConflictExpr(void)
READ_DONE();
}
static PartitionPruneStepOp *
_readPartitionPruneStepOp(void)
{
READ_LOCALS(PartitionPruneStepOp);
READ_INT_FIELD(step.step_id);
READ_INT_FIELD(opstrategy);
READ_NODE_FIELD(exprs);
READ_NODE_FIELD(cmpfns);
READ_BITMAPSET_FIELD(nullkeys);
READ_DONE();
}
static PartitionPruneStepCombine *
_readPartitionPruneStepCombine(void)
{
READ_LOCALS(PartitionPruneStepCombine);
READ_INT_FIELD(step.step_id);
READ_ENUM_FIELD(combineOp, PartitionPruneCombineOp);
READ_NODE_FIELD(source_stepids);
READ_DONE();
}
/*
* _readMergeAction
*/
@@ -2615,6 +2641,10 @@ parseNodeString(void)
return_value = _readOnConflictExpr();
else if (MATCH("MERGEACTION", 11))
return_value = _readMergeAction();
else if (MATCH("PARTITIONPRUNESTEPOP", 20))
return_value = _readPartitionPruneStepOp();
else if (MATCH("PARTITIONPRUNESTEPCOMBINE", 25))
return_value = _readPartitionPruneStepCombine();
else if (MATCH("RTE", 3))
return_value = _readRangeTblEntry();
else if (MATCH("RANGETBLFUNCTION", 16))