|
| 1 | +// <copyright file="BTreeIndexRangeQueryBenchmark.cs" company="MPCoreDeveloper"> |
| 2 | +// Copyright (c) 2025-2026 MPCoreDeveloper and GitHub Copilot. All rights reserved. |
| 3 | +// Licensed under the MIT License. See LICENSE file in the project root for full license information. |
| 4 | +// </copyright> |
| 5 | + |
| 6 | +using BenchmarkDotNet.Attributes; |
| 7 | +using BenchmarkDotNet.Order; |
| 8 | +using SharpCoreDB.Benchmarks.Infrastructure; |
| 9 | + |
| 10 | +namespace SharpCoreDB.Benchmarks; |
| 11 | + |
| 12 | +/// <summary> |
| 13 | +/// Benchmarks B-tree index performance for range queries vs full table scan. |
| 14 | +/// Target: B-tree range query < 10ms vs ~30ms full scan on 10k records. |
| 15 | +/// |
| 16 | +/// Test Scenarios: |
| 17 | +/// 1. Full Table Scan (baseline) - SELECT * WHERE age > 30 |
| 18 | +/// 2. Hash Index Scan (control) - Falls back to full scan for range queries |
| 19 | +/// 3. B-Tree Range Query - SELECT * WHERE age > 30 using B-tree index |
| 20 | +/// 4. B-Tree BETWEEN Query - SELECT * WHERE age BETWEEN 25 AND 35 |
| 21 | +/// 5. B-Tree ORDER BY - SELECT * FROM users ORDER BY age (uses B-tree sorted iteration) |
| 22 | +/// |
| 23 | +/// Expected Results: |
| 24 | +/// - Full scan: ~30ms (10k records, columnar storage) |
| 25 | +/// - Hash index: ~30ms (no benefit for range queries) |
| 26 | +/// - B-tree range: < 10ms (3-5x faster with index) |
| 27 | +/// - B-tree BETWEEN: < 15ms (selective range) |
| 28 | +/// - B-tree ORDER BY: < 5ms (sorted iteration vs external sort) |
| 29 | +/// </summary> |
| 30 | +[MemoryDiagnoser] |
| 31 | +[Orderer(SummaryOrderPolicy.FastestToSlowest)] |
| 32 | +[RankColumn] |
| 33 | +public class BTreeIndexRangeQueryBenchmark |
| 34 | +{ |
| 35 | + private BenchmarkDatabaseHelper? _dbFullScan; |
| 36 | + private BenchmarkDatabaseHelper? _dbHashIndex; |
| 37 | + private BenchmarkDatabaseHelper? _dbBTreeIndex; |
| 38 | + private const int RECORD_COUNT = 10_000; |
| 39 | + private const string DB_PATH_FULLSCAN = "./bench_btree_fullscan"; |
| 40 | + private const string DB_PATH_HASH = "./bench_btree_hash"; |
| 41 | + private const string DB_PATH_BTREE = "./bench_btree_btree"; |
| 42 | + |
| 43 | + [GlobalSetup] |
| 44 | + public void Setup() |
| 45 | + { |
| 46 | + // Clean up any existing test databases |
| 47 | + CleanupDatabase(DB_PATH_FULLSCAN); |
| 48 | + CleanupDatabase(DB_PATH_HASH); |
| 49 | + CleanupDatabase(DB_PATH_BTREE); |
| 50 | + |
| 51 | + // Setup 1: Full scan (no indexes) |
| 52 | + _dbFullScan = new BenchmarkDatabaseHelper(DB_PATH_FULLSCAN, enableEncryption: false); |
| 53 | + _dbFullScan.CreateUsersTableColumnar(); |
| 54 | + InsertTestData(_dbFullScan); |
| 55 | + |
| 56 | + // Setup 2: Hash index (for comparison - should not help with range queries) |
| 57 | + _dbHashIndex = new BenchmarkDatabaseHelper(DB_PATH_HASH, enableEncryption: false); |
| 58 | + _dbHashIndex.CreateUsersTableColumnar(); |
| 59 | + InsertTestData(_dbHashIndex); |
| 60 | + // Create hash index on age - won't help with range queries |
| 61 | + ((SharpCoreDB.Database)((dynamic)_dbHashIndex).database) |
| 62 | + .ExecuteSQL("CREATE INDEX idx_age_hash ON users (age)"); |
| 63 | + |
| 64 | + // Setup 3: B-tree index (should significantly improve range queries) |
| 65 | + _dbBTreeIndex = new BenchmarkDatabaseHelper(DB_PATH_BTREE, enableEncryption: false); |
| 66 | + _dbBTreeIndex.CreateUsersTableColumnar(); |
| 67 | + InsertTestData(_dbBTreeIndex); |
| 68 | + // Create B-tree index on age for range queries |
| 69 | + ((SharpCoreDB.Database)((dynamic)_dbBTreeIndex).database) |
| 70 | + .ExecuteSQL("CREATE INDEX idx_age_btree ON users (age) USING BTREE"); |
| 71 | + } |
| 72 | + |
| 73 | + private void InsertTestData(BenchmarkDatabaseHelper db) |
| 74 | + { |
| 75 | + // Insert 10k records with ages distributed 18-65 |
| 76 | + var users = new List<(int id, string name, string email, int age, DateTime createdAt, bool isActive)>(); |
| 77 | + |
| 78 | + for (int i = 1; i <= RECORD_COUNT; i++) |
| 79 | + { |
| 80 | + users.Add(( |
| 81 | + id: i, |
| 82 | + name: $"User{i}", |
| 83 | + email: $"user{i}@example.com", |
| 84 | + age: 18 + (i % 48), // Ages 18-65 |
| 85 | + createdAt: DateTime.UtcNow.AddDays(-i), |
| 86 | + isActive: i % 2 == 0 |
| 87 | + )); |
| 88 | + } |
| 89 | + |
| 90 | + // Use true batch insert for fast setup |
| 91 | + db.InsertUsersTrueBatch(users); |
| 92 | + } |
| 93 | + |
| 94 | + [GlobalCleanup] |
| 95 | + public void Cleanup() |
| 96 | + { |
| 97 | + _dbFullScan?.Dispose(); |
| 98 | + _dbHashIndex?.Dispose(); |
| 99 | + _dbBTreeIndex?.Dispose(); |
| 100 | + |
| 101 | + CleanupDatabase(DB_PATH_FULLSCAN); |
| 102 | + CleanupDatabase(DB_PATH_HASH); |
| 103 | + CleanupDatabase(DB_PATH_BTREE); |
| 104 | + } |
| 105 | + |
| 106 | + private static void CleanupDatabase(string path) |
| 107 | + { |
| 108 | + try |
| 109 | + { |
| 110 | + if (Directory.Exists(path)) |
| 111 | + { |
| 112 | + Directory.Delete(path, true); |
| 113 | + } |
| 114 | + } |
| 115 | + catch { /* Ignore cleanup errors */ } |
| 116 | + } |
| 117 | + |
| 118 | + // ==================== RANGE QUERY BENCHMARKS ==================== |
| 119 | + |
| 120 | + [Benchmark(Baseline = true)] |
| 121 | + public void FullTableScan_RangeQuery() |
| 122 | + { |
| 123 | + // SELECT * FROM users WHERE age > 30 |
| 124 | + // Expected: ~30ms (full scan of 10k records) |
| 125 | + var db = ((SharpCoreDB.Database)((dynamic)_dbFullScan!).database); |
| 126 | + var results = db.ExecuteQuery("SELECT * FROM users WHERE age > 30"); |
| 127 | + |
| 128 | + if (results.Count == 0) |
| 129 | + throw new InvalidOperationException("No results found - test data issue"); |
| 130 | + } |
| 131 | + |
| 132 | + [Benchmark] |
| 133 | + public void HashIndex_RangeQuery() |
| 134 | + { |
| 135 | + // SELECT * FROM users WHERE age > 30 (with hash index) |
| 136 | + // Expected: ~30ms (hash index doesn't help with range queries, falls back to scan) |
| 137 | + var db = ((SharpCoreDB.Database)((dynamic)_dbHashIndex!).database); |
| 138 | + var results = db.ExecuteQuery("SELECT * FROM users WHERE age > 30"); |
| 139 | + |
| 140 | + if (results.Count == 0) |
| 141 | + throw new InvalidOperationException("No results found - test data issue"); |
| 142 | + } |
| 143 | + |
| 144 | + [Benchmark] |
| 145 | + public void BTreeIndex_RangeQuery() |
| 146 | + { |
| 147 | + // SELECT * FROM users WHERE age > 30 (with B-tree index) |
| 148 | + // Expected: < 10ms (3-5x faster than full scan) |
| 149 | + var db = ((SharpCoreDB.Database)((dynamic)_dbBTreeIndex!).database); |
| 150 | + var results = db.ExecuteQuery("SELECT * FROM users WHERE age > 30"); |
| 151 | + |
| 152 | + if (results.Count == 0) |
| 153 | + throw new InvalidOperationException("No results found - test data issue"); |
| 154 | + } |
| 155 | + |
| 156 | + [Benchmark] |
| 157 | + public void BTreeIndex_BetweenQuery() |
| 158 | + { |
| 159 | + // SELECT * FROM users WHERE age BETWEEN 25 AND 35 |
| 160 | + // Expected: < 15ms (selective range query) |
| 161 | + var db = ((SharpCoreDB.Database)((dynamic)_dbBTreeIndex!).database); |
| 162 | + var results = db.ExecuteQuery("SELECT * FROM users WHERE age >= 25 AND age <= 35"); |
| 163 | + |
| 164 | + if (results.Count == 0) |
| 165 | + throw new InvalidOperationException("No results found - test data issue"); |
| 166 | + } |
| 167 | + |
| 168 | + [Benchmark] |
| 169 | + public void BTreeIndex_OrderBy() |
| 170 | + { |
| 171 | + // SELECT * FROM users ORDER BY age |
| 172 | + // Expected: < 5ms (B-tree provides sorted iteration, no external sort needed) |
| 173 | + var db = ((SharpCoreDB.Database)((dynamic)_dbBTreeIndex!).database); |
| 174 | + var results = db.ExecuteQuery("SELECT * FROM users ORDER BY age"); |
| 175 | + |
| 176 | + if (results.Count != RECORD_COUNT) |
| 177 | + throw new InvalidOperationException($"Expected {RECORD_COUNT} results, got {results.Count}"); |
| 178 | + } |
| 179 | + |
| 180 | + [Benchmark] |
| 181 | + public void FullTableScan_OrderBy() |
| 182 | + { |
| 183 | + // SELECT * FROM users ORDER BY age (without B-tree index) |
| 184 | + // Expected: ~40ms (full scan + external sort overhead) |
| 185 | + var db = ((SharpCoreDB.Database)((dynamic)_dbFullScan!).database); |
| 186 | + var results = db.ExecuteQuery("SELECT * FROM users ORDER BY age"); |
| 187 | + |
| 188 | + if (results.Count != RECORD_COUNT) |
| 189 | + throw new InvalidOperationException($"Expected {RECORD_COUNT} results, got {results.Count}"); |
| 190 | + } |
| 191 | + |
| 192 | + // ==================== POINT LOOKUP COMPARISON ==================== |
| 193 | + |
| 194 | + [Benchmark] |
| 195 | + public void HashIndex_PointLookup() |
| 196 | + { |
| 197 | + // SELECT * FROM users WHERE age = 30 (with hash index) |
| 198 | + // Expected: < 1ms (O(1) hash lookup) |
| 199 | + var db = ((SharpCoreDB.Database)((dynamic)_dbHashIndex!).database); |
| 200 | + var results = db.ExecuteQuery("SELECT * FROM users WHERE age = 30"); |
| 201 | + } |
| 202 | + |
| 203 | + [Benchmark] |
| 204 | + public void BTreeIndex_PointLookup() |
| 205 | + { |
| 206 | + // SELECT * FROM users WHERE age = 30 (with B-tree index) |
| 207 | + // Expected: ~2ms (O(log n) B-tree lookup, slower than hash but still fast) |
| 208 | + var db = ((SharpCoreDB.Database)((dynamic)_dbBTreeIndex!).database); |
| 209 | + var results = db.ExecuteQuery("SELECT * FROM users WHERE age = 30"); |
| 210 | + } |
| 211 | +} |
0 commit comments