Lower threshold for MapJoin for Scale

ydb/core/kqp/opt/kqp_statistics_transformer.cpp

@@ -360,42 +360,14 @@ class TKqpOlapPredicateSelectivityComputer: public TPredicateSelectivityComputer
             resSelectivity = tmpSelectivity;
         } else if (auto notNode = input.Maybe<TKqpOlapNot>()) {
             resSelectivity = 1 - Compute(notNode.Cast().Value());
-        } else if (input.Maybe<TCoAtomList>() && input.Ptr()->ChildrenSize() >= 1) {
-            auto listPtr = input.Maybe<TCoAtomList>().Cast().Ptr()->Child(1);
-            size_t listSize = listPtr->ChildrenSize();
-
-            if (listSize == 3) {
-                TString compSign = TString(listPtr->Child(0)->Content());
-                TString attr = TString(listPtr->Child(1)->Content());
-
-                TExprContext dummyCtx;
-                TPositionHandle dummyPos;
-
-                auto rowArg = 
-                    Build<TCoArgument>(dummyCtx, dummyPos)
-                        .Name("row")
-                    .Done();
-
-                auto member = 
-                        Build<TCoMember>(dummyCtx, dummyPos)
-                            .Struct(rowArg)
-                            .Name().Build(attr)
-                        .Done();
-
-                auto value = TExprBase(listPtr->ChildPtr(2));
-                if (listPtr->ChildPtr(2)->ChildrenSize() >= 2 && listPtr->ChildPtr(2)->ChildPtr(0)->Content() == "just") {
-                    value = TExprBase(listPtr->ChildPtr(2)->ChildPtr(1));
-                }
-                if (OlapCompSigns.contains(compSign)) {
-                    resSelectivity = this->ComputeComparisonSelectivity(member, value);
-                } else if (compSign == "eq") {
-                    resSelectivity = this->ComputeEqualitySelectivity(member, value);
-                } else if (compSign == "neq") {
-                    resSelectivity = 1 - this->ComputeEqualitySelectivity(member, value);
-                } else if (RegexpSigns.contains(compSign)) {
-                    return 0.5;
-                }
+        } else if (input.Maybe<TCoAtomList>()) {
+            auto list = input.Maybe<TCoAtomList>().Cast().Ptr();
+            resSelectivity = ComputeListSelectivity(list);
+
+            if (!resSelectivity.has_value() && list->ChildrenSize() >= 1) {
+                resSelectivity = ComputeListSelectivity(list->Child(1));
             }
+
         }
 
         if (!resSelectivity.has_value()) {
@@ -408,6 +380,46 @@ class TKqpOlapPredicateSelectivityComputer: public TPredicateSelectivityComputer
     }
 
 private:
+    std::optional<double> ComputeListSelectivity(const TExprNode::TPtr& listPtr) {
+        std::optional<double> resSelectivity;
+
+        size_t listSize = listPtr->ChildrenSize();
+        if (listSize == 3) {
+            TString compSign = TString(listPtr->Child(0)->Content());
+            TString attr = TString(listPtr->Child(1)->Content());
+
+            TExprContext dummyCtx;
+            TPositionHandle dummyPos;
+
+            auto rowArg = 
+                Build<TCoArgument>(dummyCtx, dummyPos)
+                    .Name("row")
+                .Done();
+
+            auto member = 
+                    Build<TCoMember>(dummyCtx, dummyPos)
+                        .Struct(rowArg)
+                        .Name().Build(attr)
+                    .Done();
+
+            auto value = TExprBase(listPtr->ChildPtr(2));
+            if (listPtr->ChildPtr(2)->ChildrenSize() >= 2 && listPtr->ChildPtr(2)->ChildPtr(0)->Content() == "just") {
+                value = TExprBase(listPtr->ChildPtr(2)->ChildPtr(1));
+            }
+            if (OlapCompSigns.contains(compSign)) {
+                resSelectivity = this->ComputeComparisonSelectivity(member, value);
+            } else if (compSign == "eq") {
+                resSelectivity = this->ComputeEqualitySelectivity(member, value);
+            } else if (compSign == "neq") {
+                resSelectivity = 1 - this->ComputeEqualitySelectivity(member, value);
+            } else if (RegexpSigns.contains(compSign)) {
+                return 0.5;
+            }
+        }
+
+        return resSelectivity;
+    }
+
     THashSet<TString> OlapCompSigns = {
         {"lt"},
         {"lte"},

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

@@ -163,7 +163,7 @@ bool TKqpProviderContext::IsJoinApplicable(const std::shared_ptr<IBaseOptimizerN
             return IsLookupJoinApplicable(right, left, joinConditions, rightJoinKeys, leftJoinKeys, *this);
 
         case EJoinAlgoType::MapJoin:
-            return joinKind != EJoinKind::OuterJoin && joinKind != EJoinKind::Exclusion && right->Stats->ByteSize < 1e8;
+            return joinKind != EJoinKind::OuterJoin && joinKind != EJoinKind::Exclusion && right->Stats->ByteSize < 1e5;
         case EJoinAlgoType::GraceJoin:
             return true;
         default:

ydb/core/kqp/ut/join/data/queries/tpch20.sql

@@ -0,0 +1,46 @@
+PRAGMA TablePathPrefix="/Root";
+
+-- TPC-H/TPC-R Potential Part Promotion Query (Q20)
+-- TPC TPC-H Parameter Substitution (Version 2.17.2 build 0)
+-- using 1680793381 as a seed to the RNG
+
+
+select
+    s_name,
+    s_address
+from
+    supplier
+    cross join nation
+    cross join (
+        select
+            ps_suppkey
+        from
+            partsupp
+            cross join part
+            cross join (
+                select
+                    l_partkey,
+                    l_suppkey,
+                    0.5 * sum(l_quantity) as q_threshold
+                from
+                    lineitem
+                where
+                    l_shipdate >= cast(date('1993-01-01') as Datetime)
+                    and l_shipdate < cast(date('1993-01-01') as Datetime) + interval('P365D')
+                group by
+                    l_partkey,
+                    l_suppkey
+            ) as threshold
+        where
+            ps_partkey = p_partkey
+            and ps_partkey = l_partkey
+            and ps_suppkey = l_suppkey
+            and p_name like 'maroon%'
+            and ps_availqty > threshold.q_threshold
+    ) as partsupp
+where
+    s_suppkey = ps_suppkey
+    and s_nationkey = n_nationkey
+    and n_name = 'VIETNAM'
+order by
+    s_name;

ydb/core/kqp/ut/join/data/schema/tpch.sql

@@ -32,7 +32,7 @@ CREATE TABLE `/Root/lineitem` (
 
 CREATE TABLE `/Root/nation` (
     n_comment String ,
-    n_name String ,
+    n_name String NOT NULL,
     n_nationkey Int32   NOT NULL, -- Identifier
     n_regionkey Int32 , -- FK to R_REGIONKEY
     PRIMARY KEY(n_nationkey)

ydb/core/kqp/ut/join/data/stats/tpch100s_no_countmin.json

@@ -0,0 +1,68 @@
+{
+    "/Root/customer": {
+        "byte_size": 1578400481,
+        "n_rows": 15000000,
+        "n_attrs": 8,
+        "key_columns": [
+            "c_custkey"
+        ]
+    },
+    "/Root/lineitem": {
+        "byte_size": 41682657607,
+        "n_rows": 600037902,
+        "n_attrs": 16,
+        "key_columns": [
+            "l_linenumber",
+            "l_orderkey"
+        ]
+    },
+    "/Root/nation": {
+        "byte_size": 3758,
+        "n_rows": 25,
+        "n_attrs": 4,
+        "key_columns": [
+            "n_nationkey"
+        ]
+    },
+    "/Root/orders": {
+        "byte_size": 9159017758,
+        "n_rows": 150000000,
+        "n_attrs": 9,
+        "key_columns": [
+            "o_orderkey"
+        ]
+    },
+    "/Root/part": {
+        "byte_size": 1460288880,
+        "n_rows": 20000000,
+        "n_attrs": 9,
+        "key_columns": [
+            "p_partkey"
+        ]
+    },
+    "/Root/partsupp": {
+        "byte_size": 5068061409,
+        "n_rows": 80000000,
+        "n_attrs": 5,
+        "key_columns": [
+            "ps_partkey",
+            "ps_suppkey"
+        ]
+    },
+    "/Root/region": {
+        "byte_size": 1280,
+        "n_rows": 5,
+        "n_attrs": 3,
+        "key_columns": [
+            "r_regionkey"
+        ]
+    },
+    "/Root/supplier": {
+        "byte_size": 95577702,
+        "n_rows": 1000000,
+        "n_attrs": 7,
+        "key_columns": [
+            "s_suppkey"
+        ]
+    }
+}

ydb/core/kqp/ut/join/kqp_join_order_ut.cpp

@@ -450,6 +450,10 @@ Y_UNIT_TEST_SUITE(KqpJoinOrder) {
         ExecuteJoinOrderTestDataQueryWithStats("queries/tpch11.sql", "stats/tpch1000s.json", StreamLookupJoin, ColumnStore);
     }
 
+    Y_UNIT_TEST_XOR_OR_BOTH_FALSE(TPCH20, StreamLookupJoin, ColumnStore) {
+        ExecuteJoinOrderTestDataQueryWithStats("queries/tpch20.sql", "stats/tpch100s_no_countmin.json", StreamLookupJoin, ColumnStore);
+    }
+
     Y_UNIT_TEST_XOR_OR_BOTH_FALSE(TPCH21, StreamLookupJoin, ColumnStore) {
         ExecuteJoinOrderTestDataQueryWithStats("queries/tpch21.sql", "stats/tpch1000s.json", StreamLookupJoin, ColumnStore);
     }