Replace BwTree with BPlusTree (cmu-db#1506)

* Replacing BwTree with BPlusTree * Format * Change in postgres parser * Remove BwTree test in tpcc_test Co-authored-by: Matt Butrovich <[email protected]> Co-authored-by: Lin Ma <[email protected]>
ZhangJieee · Mar 11, 2021 · af3816b · af3816b
1 parent a34360a
commit af3816b
Show file tree

Hide file tree

Showing 22 changed files with 92 additions and 85 deletions.
diff --git a/benchmark/catalog/catalog_benchmark.cpp b/benchmark/catalog/catalog_benchmark.cpp
@@ -105,7 +105,7 @@ class CatalogBenchmark : public benchmark::Fixture {
                                 const catalog::Schema::Column &col) {
     std::vector<catalog::IndexSchema::Column> key_cols{catalog::IndexSchema::Column{
         col.Name(), type::TypeId::INTEGER, false, parser::ColumnValueExpression(db_, table_oid, col.Oid())}};
-    auto index_schema = catalog::IndexSchema(key_cols, storage::index::IndexType::BWTREE, true, true, false, true);
+    auto index_schema = catalog::IndexSchema(key_cols, storage::index::IndexType::BPLUSTREE, true, true, false, true);
     const auto idx_oid = accessor->CreateIndex(accessor->GetDefaultNamespace(), table_oid, index_name, index_schema);
     NOISEPAGE_ASSERT(idx_oid != catalog::INVALID_INDEX_OID, "index creation should not fail");
     auto true_schema = accessor->GetIndexSchema(idx_oid);

diff --git a/benchmark/storage/index_wrapper_benchmark.cpp b/benchmark/storage/index_wrapper_benchmark.cpp
@@ -46,7 +46,7 @@ class IndexBenchmark : public benchmark::Fixture {
   storage::ProjectedRowInitializer tuple_initializer_ =
       storage::ProjectedRowInitializer::Create(std::vector<uint16_t>{1}, std::vector<uint16_t>{1});  // This is a dummy
 
-  // HashIndex or BwTreeIndex
+  // HashIndex or BPlusTreeIndex
   common::ManagedPointer<storage::index::Index> index_;
   transaction::TimestampManager *timestamp_manager_;
   transaction::DeferredActionManager *deferred_action_manager_;
@@ -116,7 +116,7 @@ class IndexBenchmark : public benchmark::Fixture {
   }
 
   // Table is populated with sequentially increasing keys, which will not make performance difference on hash index, but
-  // may on the BwTree structure
+  // may on the BPlusTree structure
   void PopulateTableAndIndex() {
     // Set up keystore and transaction manager for insert actions
     auto *const insert_key = index_->GetProjectedRowInitializer().InitializeRow(key_buffer_);
@@ -168,11 +168,11 @@ class IndexBenchmark : public benchmark::Fixture {
   }
 };
 
-// Determine required time to run key lookup with BwTree structure for index
+// Determine required time to run key lookup with BplusTree structure for index
 // NOLINTNEXTLINE
-BENCHMARK_DEFINE_F(IndexBenchmark, BwTreeIndexRandomScanKey)(benchmark::State &state) {
-  // Create index using BwTree and populate associated table
-  CreateIndex(storage::index::IndexType::BWTREE);
+BENCHMARK_DEFINE_F(IndexBenchmark, BPlusTreeIndexRandomScanKey)(benchmark::State &state) {
+  // Create index using BPlusTree and populate associated table
+  CreateIndex(storage::index::IndexType::BPLUSTREE);
   PopulateTableAndIndex();
   // NOLINTNEXTLINE
   for (auto _ : state) {
@@ -203,7 +203,7 @@ BENCHMARK_DEFINE_F(IndexBenchmark, HashIndexRandomScanKey)(benchmark::State &sta
 // BENCHMARK REGISTRATION
 // ----------------------------------------------------------------------------
 // clang-format off
-BENCHMARK_REGISTER_F(IndexBenchmark, BwTreeIndexRandomScanKey)
+BENCHMARK_REGISTER_F(IndexBenchmark, BPlusTreeIndexRandomScanKey)
     ->UseManualTime()
     ->Unit(benchmark::kMillisecond);
 BENCHMARK_REGISTER_F(IndexBenchmark, HashIndexRandomScanKey)

diff --git a/benchmark/storage/recovery_benchmark.cpp b/benchmark/storage/recovery_benchmark.cpp
@@ -181,7 +181,7 @@ BENCHMARK_DEFINE_F(RecoveryBenchmark, IndexRecovery)(benchmark::State &state) {
       auto index_col =
           catalog::IndexSchema::Column("index_col", type::TypeId::INTEGER, false,
                                        parser::ColumnValueExpression(db_oid, table_oid, schema.GetColumn(0).Oid()));
-      catalog::IndexSchema index_schema({index_col}, storage::index::IndexType::BWTREE, true, false, false, true);
+      catalog::IndexSchema index_schema({index_col}, storage::index::IndexType::BPLUSTREE, true, false, false, true);
       auto index_oid =
           catalog_accessor->CreateIndex(namespace_oid, table_oid, index_name + std::to_string(i), index_schema);
       auto *index = storage::index::IndexBuilder().SetKeySchema(index_schema).Build();

diff --git a/src/catalog/postgres/builder.cpp b/src/catalog/postgres/builder.cpp
@@ -495,8 +495,8 @@ IndexSchema Builder::GetColumnOidIndexSchema(db_oid_t db) {
                        parser::ColumnValueExpression(db, PgAttribute::COLUMN_TABLE_OID, PgAttribute::ATTNUM.oid_));
   columns.back().SetOid(indexkeycol_oid_t(2));
 
-  // Primary, must be a BWTREE due to ScanAscending usage
-  IndexSchema schema(columns, storage::index::IndexType::BWTREE, true, true, false, true);
+  // Primary, must be a BPLUSTREE due to ScanAscending usage
+  IndexSchema schema(columns, storage::index::IndexType::BPLUSTREE, true, true, false, true);
 
   return schema;
 }
@@ -830,7 +830,7 @@ IndexSchema Builder::GetProcNameIndexSchema(db_oid_t db) {
   columns.back().SetOid(indexkeycol_oid_t(2));
 
   // Non-Unique, not primary
-  IndexSchema schema(columns, storage::index::IndexType::BWTREE, false, false, false, false);
+  IndexSchema schema(columns, storage::index::IndexType::BPLUSTREE, false, false, false, false);
 
   return schema;
 }
@@ -848,7 +848,7 @@ IndexSchema Builder::GetStatisticOidIndexSchema(db_oid_t db) {
   columns.back().SetOid(indexkeycol_oid_t(2));
 
   // Primary
-  IndexSchema schema(columns, storage::index::IndexType::BWTREE, true, true, false, true);
+  IndexSchema schema(columns, storage::index::IndexType::BPLUSTREE, true, true, false, true);
 
   return schema;
 }

diff --git a/src/execution/sql/ddl_executors.cpp b/src/execution/sql/ddl_executors.cpp
@@ -65,7 +65,7 @@ bool DDLExecutors::CreateTableExecutor(const common::ManagedPointer<planner::Cre
                               parser::ColumnValueExpression(connection_db, table_oid, table_col.Oid()));
       }
     }
-    catalog::IndexSchema index_schema(key_cols, storage::index::IndexType::BWTREE, true, true, false, true);
+    catalog::IndexSchema index_schema(key_cols, storage::index::IndexType::BPLUSTREE, true, true, false, true);
 
     // Create the index, and use its return value as overall success result
     result = result &&
@@ -87,7 +87,7 @@ bool DDLExecutors::CreateTableExecutor(const common::ManagedPointer<planner::Cre
                               parser::ColumnValueExpression(connection_db, table_oid, table_col.Oid()));
       }
     }
-    catalog::IndexSchema index_schema(key_cols, storage::index::IndexType::BWTREE, true, false, false, true);
+    catalog::IndexSchema index_schema(key_cols, storage::index::IndexType::BPLUSTREE, true, false, false, true);
 
     // Create the index, and use its return value as overall success result
     result = result && CreateIndex(accessor, node->GetNamespaceOid(), unique_constraint.constraint_name_, table_oid,

diff --git a/src/optimizer/rules/implementation_rules.cpp b/src/optimizer/rules/implementation_rules.cpp
@@ -911,7 +911,7 @@ void LogicalCreateIndexToPhysicalCreateIndex::Transform(
       cols.emplace_back(name, type, nullable, *attr);
   }
 
-  storage::index::IndexType idx_type = storage::index::IndexType::BWTREE;
+  storage::index::IndexType idx_type = storage::index::IndexType::BPLUSTREE;
   switch (ci_op->GetIndexType()) {
     case parser::IndexType::BWTREE:
       idx_type = storage::index::IndexType::BWTREE;

diff --git a/src/parser/postgresparser.cpp b/src/parser/postgresparser.cpp
@@ -1381,9 +1381,9 @@ std::unique_ptr<SQLStatement> PostgresParser::CreateIndexTransform(ParseResult *
   // TODO(WAN): do we need to do case conversion?
   if (strcmp(access_method, "invalid") == 0) {
     index_type = IndexType::INVALID;
-  } else if ((strcmp(access_method, "btree") == 0) || (strcmp(access_method, "bwtree") == 0)) {
+  } else if (strcmp(access_method, "bwtree") == 0) {
     index_type = IndexType::BWTREE;
-  } else if ((strcmp(access_method, "b+tree") == 0) || (strcmp(access_method, "bplustree") == 0)) {
+  } else if ((strcmp(access_method, "btree") == 0) || (strcmp(access_method, "bplustree") == 0)) {
     index_type = IndexType::BPLUSTREE;
   } else if (strcmp(access_method, "hash") == 0) {
     index_type = IndexType::HASH;

diff --git a/test/catalog/catalog_test.cpp b/test/catalog/catalog_test.cpp
@@ -324,7 +324,7 @@ TEST_F(CatalogTests, UserIndexTest) {
   // Create the index
   std::vector<catalog::IndexSchema::Column> key_cols{catalog::IndexSchema::Column{
       "id", type::TypeId::INTEGER, false, parser::ColumnValueExpression(db_, table_oid, schema.GetColumn("id").Oid())}};
-  auto index_schema = catalog::IndexSchema(key_cols, storage::index::IndexType::BWTREE, true, true, false, true);
+  auto index_schema = catalog::IndexSchema(key_cols, storage::index::IndexType::BPLUSTREE, true, true, false, true);
   auto idx_oid = accessor->CreateIndex(accessor->GetDefaultNamespace(), table_oid,
                                        "test_table_index_mabobberwithareallylongnamethatstillneedsmore", index_schema);
   EXPECT_NE(idx_oid, catalog::INVALID_INDEX_OID);
@@ -379,7 +379,7 @@ TEST_F(CatalogTests, CascadingDropTableTest) {
   EXPECT_NE(accessor, nullptr);
   std::vector<catalog::IndexSchema::Column> key_cols{catalog::IndexSchema::Column{
       "id", type::TypeId::INTEGER, false, parser::ColumnValueExpression(db_, table_oid, schema.GetColumn("id").Oid())}};
-  auto index_schema = catalog::IndexSchema(key_cols, storage::index::IndexType::BWTREE, true, true, false, true);
+  auto index_schema = catalog::IndexSchema(key_cols, storage::index::IndexType::BPLUSTREE, true, true, false, true);
   auto idx_oid = accessor->CreateIndex(accessor->GetDefaultNamespace(), table_oid, "test_index", index_schema);
   EXPECT_NE(idx_oid, catalog::INVALID_INDEX_OID);
   auto true_schema = accessor->GetIndexSchema(idx_oid);
@@ -443,7 +443,7 @@ TEST_F(CatalogTests, CascadingDropNamespaceTest) {
   EXPECT_NE(accessor, nullptr);
   std::vector<catalog::IndexSchema::Column> key_cols{catalog::IndexSchema::Column{
       "id", type::TypeId::INTEGER, false, parser::ColumnValueExpression(db_, table_oid, schema.GetColumn("id").Oid())}};
-  auto index_schema = catalog::IndexSchema(key_cols, storage::index::IndexType::BWTREE, true, true, false, true);
+  auto index_schema = catalog::IndexSchema(key_cols, storage::index::IndexType::BPLUSTREE, true, true, false, true);
   auto idx_oid = accessor->CreateIndex(ns_oid, table_oid, "test_index", index_schema);
   EXPECT_NE(idx_oid, catalog::INVALID_INDEX_OID);
   auto true_schema = accessor->GetIndexSchema(idx_oid);
@@ -508,7 +508,7 @@ TEST_F(CatalogTests, CascadingDropNamespaceWithIndexOnOtherNamespaceTest) {
   EXPECT_NE(accessor, nullptr);
   std::vector<catalog::IndexSchema::Column> key_cols{catalog::IndexSchema::Column{
       "id", type::TypeId::INTEGER, false, parser::ColumnValueExpression(db_, table_oid, schema.GetColumn("id").Oid())}};
-  auto index_schema = catalog::IndexSchema(key_cols, storage::index::IndexType::BWTREE, true, true, false, true);
+  auto index_schema = catalog::IndexSchema(key_cols, storage::index::IndexType::BPLUSTREE, true, true, false, true);
   auto idx_oid = accessor->CreateIndex(ns_oid, table_oid, "test_index", index_schema);
   EXPECT_NE(idx_oid, catalog::INVALID_INDEX_OID);
   auto true_schema = accessor->GetIndexSchema(idx_oid);
@@ -689,7 +689,7 @@ TEST_F(CatalogTests, GetIndexesTest) {
   // Create the index
   std::vector<catalog::IndexSchema::Column> key_cols{catalog::IndexSchema::Column{
       "id", type::TypeId::INTEGER, false, parser::ColumnValueExpression(db_, table_oid, schema.GetColumn("id").Oid())}};
-  auto index_schema = catalog::IndexSchema(key_cols, storage::index::IndexType::BWTREE, true, true, false, true);
+  auto index_schema = catalog::IndexSchema(key_cols, storage::index::IndexType::BPLUSTREE, true, true, false, true);
   auto idx_oid = accessor->CreateIndex(accessor->GetDefaultNamespace(), table_oid, "test_table_idx", index_schema);
   EXPECT_NE(idx_oid, catalog::INVALID_INDEX_OID);
   auto true_schema = accessor->GetIndexSchema(idx_oid);
@@ -736,7 +736,7 @@ TEST_F(CatalogTests, GetIndexObjectsTest) {
     std::vector<catalog::IndexSchema::Column> key_cols{
         catalog::IndexSchema::Column{"id", type::TypeId::INTEGER, false,
                                      parser::ColumnValueExpression(db_, table_oid, schema.GetColumn("id").Oid())}};
-    auto index_schema = catalog::IndexSchema(key_cols, storage::index::IndexType::BWTREE, true, true, false, true);
+    auto index_schema = catalog::IndexSchema(key_cols, storage::index::IndexType::BPLUSTREE, true, true, false, true);
     auto idx_oid = accessor->CreateIndex(accessor->GetDefaultNamespace(), table_oid,
                                          "test_table_idx" + std::to_string(i), index_schema);
     EXPECT_NE(idx_oid, catalog::INVALID_INDEX_OID);

diff --git a/test/execution/ddl_executors_test.cpp b/test/execution/ddl_executors_test.cpp
@@ -46,7 +46,7 @@ class DDLExecutorsTests : public TerrierTest {
                                                        catalog::col_oid_t(1)));
     StorageTestUtil::ForceOid(&(keycols[0]), catalog::indexkeycol_oid_t(1));
     index_schema_ =
-        std::make_unique<catalog::IndexSchema>(keycols, storage::index::IndexType::BWTREE, true, true, false, true);
+        std::make_unique<catalog::IndexSchema>(keycols, storage::index::IndexType::BPLUSTREE, true, true, false, true);
 
     txn_ = txn_manager_->BeginTransaction();
     accessor_ = catalog_->GetAccessor(common::ManagedPointer(txn_), db_, DISABLED);

diff --git a/test/execution/index_create_test.cpp b/test/execution/index_create_test.cpp
@@ -106,7 +106,7 @@ TEST_F(IndexCreateTest, SimpleIndexCreate) {
   parser::ColumnValueExpression col_expr(table_oid, table_col.Oid(), table_col.Type());
   index_cols.emplace_back("index_colA", type::TypeId::INTEGER, false, col_expr);
 
-  catalog::IndexSchema tmp_index_schema{index_cols, storage::index::IndexType::BWTREE, false, false, false, false};
+  catalog::IndexSchema tmp_index_schema{index_cols, storage::index::IndexType::BPLUSTREE, false, false, false, false};
 
   CreateIndex(table_oid, "indexA", std::make_unique<catalog::IndexSchema>(tmp_index_schema));
 
@@ -126,7 +126,7 @@ TEST_F(IndexCreateTest, SimpleIndexCreate2) {
   parser::ColumnValueExpression col_expr(table_oid, table_col.Oid(), table_col.Type());
   index_cols.emplace_back("index_colE", type::TypeId::INTEGER, false, col_expr);
 
-  catalog::IndexSchema tmp_index_schema{index_cols, storage::index::IndexType::BWTREE, false, false, false, false};
+  catalog::IndexSchema tmp_index_schema{index_cols, storage::index::IndexType::BPLUSTREE, false, false, false, false};
 
   CreateIndex(table_oid, "indexE", std::make_unique<catalog::IndexSchema>(tmp_index_schema));
 
@@ -150,7 +150,7 @@ TEST_F(IndexCreateTest, MultiColumnIndexCreate) {
   parser::ColumnValueExpression col_expr_b(table_oid, table_col_b.Oid(), table_col_b.Type());
   index_cols.emplace_back("index_colB", type::TypeId::INTEGER, false, col_expr_b);
 
-  catalog::IndexSchema tmp_index_schema{index_cols, storage::index::IndexType::BWTREE, false, false, false, false};
+  catalog::IndexSchema tmp_index_schema{index_cols, storage::index::IndexType::BPLUSTREE, false, false, false, false};
 
   CreateIndex(table_oid, "indexAB", std::make_unique<catalog::IndexSchema>(tmp_index_schema));
 

diff --git a/test/include/test_util/storage_test_util.h b/test/include/test_util/storage_test_util.h
@@ -516,7 +516,7 @@ class StorageTestUtil {
       ForceOid(&(key_cols.back()), key_oid);
     }
 
-    return catalog::IndexSchema(key_cols, storage::index::IndexType::BWTREE, false, false, false, true);
+    return catalog::IndexSchema(key_cols, storage::index::IndexType::BPLUSTREE, false, false, false, true);
   }
 
   /**
@@ -555,7 +555,7 @@ class StorageTestUtil {
       bytes_used = static_cast<uint16_t>(bytes_used + type::TypeUtil::GetTypeSize(type));
     }
 
-    return catalog::IndexSchema(key_cols, storage::index::IndexType::BWTREE, false, false, false, true);
+    return catalog::IndexSchema(key_cols, storage::index::IndexType::BPLUSTREE, false, false, false, true);
   }
 
  private:

diff --git a/test/integration/tpcc_test.cpp b/test/integration/tpcc_test.cpp
@@ -104,7 +104,7 @@ class TPCCTests : public TerrierTest {
 TEST_F(TPCCTests, WithoutLoggingHashIndexes) { RunTPCC(false, false, storage::index::IndexType::HASHMAP); }
 
 // NOLINTNEXTLINE
-TEST_F(TPCCTests, WithoutLoggingBwTreeIndexes) { RunTPCC(false, false, storage::index::IndexType::BWTREE); }
+TEST_F(TPCCTests, WithoutLoggingBPlusTreeIndexes) { RunTPCC(false, false, storage::index::IndexType::BPLUSTREE); }
 
 // NOLINTNEXTLINE
 TEST_F(TPCCTests, WithLogging) { RunTPCC(true, false, storage::index::IndexType::HASHMAP); }