pluggable optimizations

src/backend/distributed/planner/path_based_planner.c

@@ -31,8 +31,6 @@
 #include "utils/builtins.h"
 #include "utils/syscache.h"
 
-typedef List *(*optimizeFn)(PlannerInfo *root, Path *originalPath);
-
 typedef struct DistributedUnionPath
 {
 	CustomPath custom_path;
@@ -136,20 +134,22 @@ Cost RepartitionStartupCost = 1000;
 Cost RepartitionPerRowCost = .000;
 Cost RepartitionPerMBCost = .01;
 
-/* list of functions that will be called to optimized in the joinhook*/
+/* list of functions that will be called to optimized in the join hook */
-static optimizeFn joinOptimizations[] = {
+OptimizationEntry joinOptimizations[] = {
-	OptimizeJoinPath,
+	{.name = "pushdown", .fn = OptimizeJoinPath, .enabled = true},
-	OptimizeRepartitionInnerJoinPath,
+	{.name = "repartition", .fn = OptimizeRepartitionInnerJoinPath, .enabled = true},
 
-/*	BroadcastOuterJoinPath, */
+	{.name = "broadcast_outer", .fn = BroadcastOuterJoinPath, .enabled = false},
-/*	BroadcastInnerJoinPath, */
+	{.name = "broadcast_inner", .fn = BroadcastInnerJoinPath, .enabled = false},
-/*	GeoOverlapJoin, */
+	{.name = "geooverlap", .fn = GeoOverlapJoin, .enabled = false},
+	{},
 };
 
-static optimizeFn groupOptimizations[] = {
+OptimizationEntry groupOptimizations[] = {
-		PushDownAggPath,
+	{.name = "pushdown", .fn = PushDownAggPath, .enabled = true},
-		PartialPushDownAggPath,
+	{.name = "partial_pushdown", .fn = PartialPushDownAggPath, .enabled = true},
-		RepartitionAggPath,
+	{.name = "repartition", .fn = RepartitionAggPath, .enabled = true},
+	{},
 };
 
 const CustomPathMethods geoScanMethods = {
@@ -1600,9 +1600,15 @@ PathBasedPlannerJoinHook(PlannerInfo *root,
 	Path *originalPath = NULL;
 	foreach_ptr(originalPath, joinrel->pathlist)
 	{
-		for (int i = 0; i < sizeof(joinOptimizations) / sizeof(joinOptimizations[1]); i++)
+		for (int i = 0; joinOptimizations[i].fn != NULL; i++)
 		{
-			List *alternativePaths = joinOptimizations[i](root, originalPath);
+			if (!joinOptimizations[i].enabled)
+			{
+				/* skip disabled optimizations */
+				continue;
+			}
+
+			List *alternativePaths = joinOptimizations[i].fn(root, originalPath);
 			newPaths = list_concat(newPaths, alternativePaths);
 		}
 	}
@@ -1998,9 +2004,15 @@ PathBasedPlannerGroupAgg(PlannerInfo *root,
 	foreach_ptr(originalPath, output_rel->pathlist)
 	{
 		/* apply all optimizations on every available path */
-		for (int i = 0; i < sizeof(groupOptimizations) / sizeof(groupOptimizations[1]); i++)
+		for (int i = 0; groupOptimizations[i].fn != NULL; i++)
 		{
-			List *alternativePaths = groupOptimizations[i](root, originalPath);
+			if (!groupOptimizations[i].enabled)
+			{
+				/* skip optimizations when they are disabled */
+				continue;
+			}
+
+			List *alternativePaths = groupOptimizations[i].fn(root, originalPath);
 			newPaths = list_concat(newPaths, alternativePaths);
 		}
 	}

src/backend/distributed/shared_library_init.c

@@ -1260,6 +1260,40 @@ RegisterCitusConfigVariables(void)
 		GUC_STANDARD,
 		NULL,NULL,NULL);
 
+	/* pathbased planner optimization rules */
+	StringInfoData buf = { 0 };
+	initStringInfo(&buf);
+
+	for (int i = 0; joinOptimizations[i].fn != NULL; i++)
+	{
+		resetStringInfo(&buf);
+		appendStringInfo(&buf, "citus.path_based_planner_join_%s", joinOptimizations[i].name);
+		DefineCustomBoolVariable(
+				pstrdup(buf.data),
+				gettext_noop("Path based planner join optimization"),
+				NULL,
+				&joinOptimizations[i].enabled,
+				joinOptimizations[i].enabled,
+				PGC_USERSET,
+				GUC_STANDARD,
+				NULL, NULL, NULL);
+	}
+
+	for (int i = 0; groupOptimizations[i].fn != NULL; i++)
+	{
+		resetStringInfo(&buf);
+		appendStringInfo(&buf, "citus.path_based_planner_group_%s", groupOptimizations[i].name);
+		DefineCustomBoolVariable(
+				pstrdup(buf.data),
+				gettext_noop("Path based planner join optimization"),
+				NULL,
+				&groupOptimizations[i].enabled,
+				groupOptimizations[i].enabled,
+				PGC_USERSET,
+				GUC_STANDARD,
+				NULL, NULL, NULL);
+	}
+
 	DefineCustomIntVariable(
 		"citus.local_shared_pool_size",
 		gettext_noop(

src/include/distributed/path_based_planner.h

@@ -8,6 +8,15 @@
 #include "nodes/parsenodes.h"
 #include "nodes/pathnodes.h"
 
+typedef List *(*optimizeFn)(PlannerInfo *root, Path *originalPath);
+
+typedef struct OptimizationEntry
+{
+	const char *name;
+	bool enabled;
+	optimizeFn fn;
+} OptimizationEntry;
+
 extern bool EnableBroadcastJoin;
 
 extern Cost CollectStartupCost;
@@ -18,6 +27,9 @@ extern Cost RepartitionStartupCost;
 extern Cost RepartitionPerRowCost;
 extern Cost RepartitionPerMBCost;
 
+extern OptimizationEntry joinOptimizations[];
+extern OptimizationEntry groupOptimizations[];
+
 extern void PathBasedPlannerRelationHook(PlannerInfo *root,
 										 RelOptInfo *relOptInfo,
 										 Index restrictionIndex,