fix(cbo): deprecate using EHashJoinMode::Map for full and exclusion joins (#6287)

Author: ulya-sidorina

ydb/core/kqp/opt/logical/kqp_opt_cbo.cpp

@@ -150,7 +150,8 @@ bool TKqpProviderContext::IsJoinApplicable(const std::shared_ptr<IBaseOptimizerN
     const std::set<std::pair<NDq::TJoinColumn, NDq::TJoinColumn>>& joinConditions,
     const TVector<TString>& leftJoinKeys,
     const TVector<TString>& rightJoinKeys,
-    EJoinAlgoType joinAlgo)  {
+    EJoinAlgoType joinAlgo,
+    EJoinKind joinKind)  {
 
     switch( joinAlgo ) {
         case EJoinAlgoType::LookupJoin:
@@ -160,7 +161,7 @@ bool TKqpProviderContext::IsJoinApplicable(const std::shared_ptr<IBaseOptimizerN
             return IsLookupJoinApplicable(left, right, joinConditions, leftJoinKeys, rightJoinKeys, *this);
 
         case EJoinAlgoType::MapJoin:
-            return right->Stats->ByteSize < 5e8;
+            return joinKind != EJoinKind::OuterJoin && joinKind != EJoinKind::Exclusion && right->Stats->ByteSize < 5e8;
         case EJoinAlgoType::GraceJoin:
             return true;
         default:

ydb/core/kqp/opt/logical/kqp_opt_cbo.h

@@ -27,7 +27,7 @@ struct TKqpProviderContext : public NYql::TBaseProviderContext {
         const std::shared_ptr<NYql::IBaseOptimizerNode>& right, 
         const std::set<std::pair<NYql::NDq::TJoinColumn, NYql::NDq::TJoinColumn>>& joinConditions,
         const TVector<TString>& leftJoinKeys, const TVector<TString>& rightJoinKeys,
-        NYql::EJoinAlgoType joinAlgo) override;
+        NYql::EJoinAlgoType joinAlgo,  NYql::EJoinKind joinKind) override;
 
     virtual double ComputeJoinCost(const NYql::TOptimizerStatistics& leftStats, const NYql::TOptimizerStatistics& rightStats, const double outputRows, const double outputByteSize, NYql::EJoinAlgoType joinAlgo) const override;
 

ydb/library/yql/core/cbo/cbo_optimizer_new.cpp

@@ -131,13 +131,15 @@ bool TBaseProviderContext::IsJoinApplicable(const std::shared_ptr<IBaseOptimizer
     const std::set<std::pair<NDq::TJoinColumn, NDq::TJoinColumn>>& joinConditions,
     const TVector<TString>& leftJoinKeys,
     const TVector<TString>& rightJoinKeys,
-    EJoinAlgoType joinAlgo) {
+    EJoinAlgoType joinAlgo,
+    EJoinKind joinKind) {
 
     Y_UNUSED(left);
     Y_UNUSED(right);
     Y_UNUSED(joinConditions);
     Y_UNUSED(leftJoinKeys);
     Y_UNUSED(rightJoinKeys);
+    Y_UNUSED(joinKind);
 
     return joinAlgo == EJoinAlgoType::MapJoin;
 }

ydb/library/yql/core/cbo/cbo_optimizer_new.h

@@ -88,7 +88,8 @@ struct IProviderContext {
         const std::set<std::pair<NDq::TJoinColumn, NDq::TJoinColumn>>& joinConditions,
         const TVector<TString>& leftJoinKeys,
         const TVector<TString>& rightJoinKeys,
-        EJoinAlgoType joinAlgo) = 0;
+        EJoinAlgoType joinAlgo,
+        EJoinKind joinKind) = 0;
 };
 
 /**
@@ -105,7 +106,8 @@ struct TBaseProviderContext : public IProviderContext {
         const std::set<std::pair<NDq::TJoinColumn, NDq::TJoinColumn>>& joinConditions,
         const TVector<TString>& leftJoinKeys,
         const TVector<TString>& rightJoinKeys,
-        EJoinAlgoType joinAlgo) override;
+        EJoinAlgoType joinAlgo,
+        EJoinKind joinKind) override;
 
     virtual TOptimizerStatistics ComputeJoinStats(
         const TOptimizerStatistics& leftStats,

ydb/library/yql/dq/opt/dq_opt_dphyp_solver.h

@@ -404,7 +404,7 @@ template <typename TNodeSet> std::shared_ptr<TJoinOptimizerNodeInternal> TDPHypS
     bool bestJoinIsReversed = false;
 
     for (auto joinAlgo : AllJoinAlgos) {
-        if (ctx.IsJoinApplicable(left, right, joinConditions, leftJoinKeys, rightJoinKeys, joinAlgo)){
+        if (ctx.IsJoinApplicable(left, right, joinConditions, leftJoinKeys, rightJoinKeys, joinAlgo, joinKind)){
             auto cost = ctx.ComputeJoinStats(*left->Stats, *right->Stats, leftJoinKeys, rightJoinKeys, joinAlgo, joinKind).Cost;
             if (cost < bestCost) {
                 bestCost = cost;
@@ -414,7 +414,7 @@ template <typename TNodeSet> std::shared_ptr<TJoinOptimizerNodeInternal> TDPHypS
         }
 
         if (isCommutative) {
-            if (ctx.IsJoinApplicable(right, left, reversedJoinConditions, rightJoinKeys, leftJoinKeys, joinAlgo)){
+            if (ctx.IsJoinApplicable(right, left, reversedJoinConditions, rightJoinKeys, leftJoinKeys, joinAlgo, joinKind)){
                 auto cost = ctx.ComputeJoinStats(*right->Stats, *left->Stats,  rightJoinKeys, leftJoinKeys, joinAlgo, joinKind).Cost;
                 if (cost < bestCost) {
                     bestCost = cost;

ydb/library/yql/dq/opt/dq_opt_phy.cpp

@@ -2666,7 +2666,8 @@ TExprBase DqBuildJoin(const TExprBase& node, TExprContext& ctx, IOptimizationCon
     const bool rightIsUnionAll = join.RightInput().Maybe<TDqCnUnionAll>().IsValid();
 
     auto joinAlgo = FromString<EJoinAlgoType>(join.JoinAlgo().StringValue());
-    if (joinAlgo == EJoinAlgoType::MapJoin) {
+    const bool mapJoinCanBeApplied = joinType != "Full"sv && joinType != "Exclusion"sv;
+    if (joinAlgo == EJoinAlgoType::MapJoin && mapJoinCanBeApplied) {
         hashJoin = EHashJoinMode::Map;
     } else if (joinAlgo == EJoinAlgoType::GraceJoin) {
         hashJoin = EHashJoinMode::GraceAndSelf;