2026-09-26 10:20:24 +08:00
2 changed files with 319 additions and 0 deletions
@@ -0,0 +1,70 @@
+# 規模效能基準：原生 postmeta vs Hot 表 vs Entity 表
+
+**量測日期**：2026-09-25　**分支**：`hardening/2.0`　**目標規模**：vendor 100,000、listing 1,000,000
+
+## 如何重現
+
+```bash
+# DB 帳密沿用整合測試的 TMDO_TEST_DB_* 環境變數；資料庫名稱必須以 _bench 結尾
+TMDO_TEST_DB_NAME=wp_wpdo_bench php tests/bench/seed.php --vendors=100000 --listings=1000000   # 約 7.5 分鐘
+TMDO_TEST_DB_NAME=wp_wpdo_bench php tests/bench/run.php --iterations=20                          # 約 20 分鐘
+```
+
+## 環境與資料
+
+| 項目 | 值 |
+|---|---|
+| 資料庫 | MariaDB 10.11.14（本機，InnoDB 預設設定） |
+| 主機 | 8 核、31 GB RAM、NVMe |
+| 資料 | 100,000 vendor（users + hp_vendor posts）、1,000,000 listing（90% publish） |
+| 原生 meta | postmeta 9,549,796 筆（951 MB）、usermeta 600,000 筆 |
+| Entity 表 | `wpdo_post_hp_listing_core` 1,000,000 筆（379 MB） |
+| Hot 表 | `wpdo_hot_hp_listing` 1,000,000 筆（229 MB） |
+
+SQL 的形狀與實際程式一致：原生是 WP_Query 的 meta_query（每個條件一個 INNER JOIN、數值用 CAST、`SQL_CALC_FOUND_ROWS`），Hot 是 `TMDO_Query_Router`，Entity 是 `TMDO_Post_Query_Router`。每個情境 20 次，另有 3 次暖機不計入，參數隨機但可重現（`--seed=7`）。寫入在交易內執行後 rollback。
+
+> 20 次取樣下 p95 等於最大值，僅供參考。
+
+## 結果
+
+| 情境 | 形式 | p50 ms | p95 ms | 結果筆數 | 使用索引 |
+|---|---|---:|---:|---:|---|
+| 1. 價格區間搜尋（命中約 20%） | 原生 | **14,695** | 14,890 | 180,292 | postmeta:post_id |
+| | Hot | 558 | 3,605 | 180,329 | h:ALL |
+| | Entity | 652 | 3,817 | 180,300 | e:ALL |
+| 2. 價格區間 + verified=1 | 原生 | **17,094** | 17,250 | 54,197 | postmeta:post_id ×2 |
+| | Hot | 277 | 299 | 54,109 | `idx_hp_verified_hp_price`（複合） |
+| | Entity | 337 | 340 | 54,180 | e:ALL |
+| 3. 某 vendor 名下的 listing | `post_parent`（HivePress 實際做法） | 0.19 | 0.25 | 7 | posts:post_parent |
+| | 原生 meta `hp_vendor` | **14,432** | 14,599 | 8 | postmeta:post_id |
+| | Hot | **3,405** | 3,651 | 10 | 無可用索引 |
+| | Entity | 0.28 | 0.33 | 8 | `idx_hp_vendor` |
+| 4. 單筆 listing 讀全部 meta | 原生 | 0.12 | 0.15 | 9 列 | postmeta:post_id |
+| | Hot / Entity | 0.09 | 0.11–0.14 | 1 列 | 主鍵 |
+| 5. 列表頁預載 20 筆 meta | 原生 | 0.59 | 0.61 | 181 列 | postmeta:post_id |
+| | Hot / Entity | 0.21–0.22 | 0.23–0.26 | 20 列 | 主鍵 |
+| 6. 單一 vendor 讀使用者 meta | 原生 | 0.11 | 0.14 | 6 列 | usermeta:user_id |
+| | Entity | 0.08 | 0.11 | 1 列 | 主鍵 |
+| 7. 寫入：更新一筆價格 | 原生 | 0.12 | 0.16 | — | — |
+| | Entity | 0.10 | 0.13 | — | — |
+| | dual_write（兩邊都寫） | 0.22 | 0.26 | — | — |
+
+## 結論
+
+1. **原生 meta_query 在百萬規模下無法使用。** 任何以 meta 為條件的搜尋都要 14–17 秒，扁平化是必要的，不是錦上添花。
+2. **扁平表的價值集中在「搜尋與篩選」。** 單筆或批次讀 meta（情境 4–6）三種形式都在 1 ms 內，差距可以忽略，而且實務上還有 object cache。
+3. **寬範圍搜尋仍有瓶頸（情境 1，p50 約 0.6 秒、p95 約 3.7 秒）。** 命中 18 萬筆時，優化器選擇全表掃描；再加上 `SQL_CALC_FOUND_ROWS` 要算總數、`ORDER BY post_date` 要排序，還要 JOIN 回 `posts` 過濾狀態。p50 和 p95 差很多，是因為不同的價格區間讓優化器選了不同的執行計畫。
+4. **Hot 表的索引規則有漏洞。** `add_covering_indexes()` 只替 decimal、tinyint，以及名稱以 `_time` 結尾的 bigint 建索引，所以 `hp_vendor` 沒有索引，情境 3 要 3.4 秒，比 Entity 慢約 12,000 倍。
+5. **Entity 表只有單欄索引。** 情境 2 輸給有複合索引的 Hot 表（337 ms 對 277 ms）。
+6. **dual_write 的寫入成本約為兩倍（0.22 ms 對 0.12 ms）**，絕對值很小，不是瓶頸。
+7. **Hot 表和 Entity 表儲存了同一份 listing 資料**（合計約 608 MB），兩者功能重疊。
+
+## 優化候選（待使用者確認後排入 backlog）
+
+| # | 候選 | 針對 | 預期效果 |
+|---|---|---|---|
+| O-1 | Entity 表把 `post_status`、`post_date` 一併存進來，並建 `(hp_status, post_date)` 等複合索引，讓搜尋不必 JOIN 回 `posts` 排序 | 情境 1、2 | 寬範圍搜尋改為走索引排序，避免 filesort |
+| O-2 | Entity 表支援群組層級的複合索引宣告，例如 `(hp_verified, hp_price)`、`(hp_featured, hp_price)` | 情境 2 | 追上或超過 Hot 表 |
+| O-3 | Hot 表的索引規則補上外鍵型 bigint 欄位（例如 `hp_vendor`） | 情境 3 | 3.4 秒降到 1 ms 以下 |
+| O-4 | 提供總數快取，或讓路由後的查詢可以不跑 `SQL_CALC_FOUND_ROWS`（MySQL 8 已棄用） | 情境 1、2 | 省去每次搜尋都要算出全部符合列的成本 |
+| O-5 | 評估 Hot 表與 Entity 表的整併，避免同一份資料存兩份 | 容量、寫入 | 省下約 229 MB 並減少一次寫入 |
@@ -0,0 +1,249 @@
+<?php
+/**
+ * Scale benchmark: native postmeta vs hot table vs entity table.
+ *
+ * Runs against the dataset from tests/bench/seed.php:
+ *   TMDO_TEST_DB_NAME=wp_wpdo_bench php tests/bench/run.php [--iterations=20] [--seed=7]
+ *
+ * SQL mirrors what the code under test emits: WP_Query meta_query (INNER JOIN
+ * per clause, CAST for numerics, SQL_CALC_FOUND_ROWS + FOUND_ROWS()),
+ * TMDO_Query_Router (hot) and TMDO_Post_Query_Router (entity, %s-quoted values).
+ * Writes run inside a transaction that is rolled back, so the dataset never changes.
+ *
+ * Output: a Markdown table on stdout (p50 / p95 / max in ms, result rows, index used).
+ *
+ * @package TMDO\Tests\Bench
+ */
+
+declare(strict_types=1);
+
+if ( 'cli' !== PHP_SAPI ) {
+	exit( 1 );
+}
+
+require_once __DIR__ . '/class-tmdo-bench-seeder.php';
+
+const P = 'wp_itest_';
+
+$opts       = getopt( '', array( 'iterations::', 'seed::' ) );
+$iterations = max( 1, (int) ( $opts['iterations'] ?? 20 ) );
+$warmup     = 3;
+$db_name    = (string) getenv( 'TMDO_TEST_DB_NAME' );
+
+try {
+	TMDO_Bench_Seeder::assert_bench_db( $db_name );
+} catch ( InvalidArgumentException $e ) {
+	fwrite( STDERR, $e->getMessage() . PHP_EOL );
+	exit( 2 );
+}
+
+$db = new mysqli(
+	getenv( 'TMDO_TEST_DB_HOST' ) ?: '127.0.0.1',
+	getenv( 'TMDO_TEST_DB_USER' ) ?: 'dbo',
+	(string) getenv( 'TMDO_TEST_DB_PASS' ),
+	$db_name
+);
+$db->set_charset( 'utf8mb4' );
+$db->query( 'SET SESSION max_statement_time = 30' ); // MariaDB: cap runaway native queries.
+
+$vendors  = (int) $db->query( 'SELECT COUNT(*) FROM `' . P . "posts` WHERE post_type = 'hp_vendor'" )->fetch_row()[0];
+$listings = (int) $db->query( 'SELECT COUNT(*) FROM `' . P . "posts` WHERE post_type = 'hp_listing'" )->fetch_row()[0];
+if ( 0 === $vendors || 0 === $listings ) {
+	fwrite( STDERR, "No bench data — run tests/bench/seed.php first.\n" );
+	exit( 3 );
+}
+
+mt_srand( (int) ( $opts['seed'] ?? 7 ) );
+
+// ── SQL builders ──────────────────────────────────────────────────────────
+
+/** WP_Query shape with one INNER JOIN per meta clause. */
+function native_search( array $meta_where ): string {
+	$joins = '';
+	$where = array();
+	foreach ( $meta_where as $n => $cond ) {
+		$alias  = 0 === $n ? P . 'postmeta' : "mt{$n}";
+		$joins .= ' INNER JOIN `' . P . 'postmeta`' . ( 0 === $n ? '' : " AS {$alias}" ) . " ON ( p.ID = {$alias}.post_id )";
+		$where[] = '( ' . sprintf( $cond, $alias ) . ' )';
+	}
+	return 'SELECT SQL_CALC_FOUND_ROWS p.ID FROM `' . P . "posts` p{$joins} WHERE 1=1 AND ( " . implode( ' AND ', $where )
+		. " ) AND p.post_type = 'hp_listing' AND ((p.post_status = 'publish')) GROUP BY p.ID ORDER BY p.post_date DESC LIMIT 0, 20";
+}
+
+/** TMDO router shape: one LEFT JOIN to a flat table plus flat WHERE clauses. */
+function flat_search( string $table, string $alias, string $conditions ): string {
+	return 'SELECT SQL_CALC_FOUND_ROWS p.ID FROM `' . P . "posts` p LEFT JOIN `" . P . "{$table}` AS `{$alias}` ON (p.ID = `{$alias}`.`post_id`)"
+		. " WHERE 1=1 AND p.post_type = 'hp_listing' AND ((p.post_status = 'publish')){$conditions} GROUP BY p.ID ORDER BY p.post_date DESC LIMIT 0, 20";
+}
+
+function random_ids( int $count, int $offset, int $max ): string {
+	$ids = array();
+	for ( $i = 0; $i < $count; $i++ ) {
+		$ids[] = $offset + mt_rand( 1, $max );
+	}
+	return implode( ',', $ids );
+}
+
+// ── Scenarios: label => [ variant => fn(): string[] (statements) ] ─────────
+
+$price = static function (): array {
+	$lo = mt_rand( 100, 80000 );
+	return array( $lo, $lo + 20000 ); // ~20% of the 100..100000 range.
+};
+
+$scenarios = array(
+	'1. 價格區間搜尋（約 20%）'           => array(
+		'native' => function () use ( $price ) {
+			[ $lo, $hi ] = $price();
+			return array( native_search( array( "%1\$s.meta_key = 'hp_price' AND CAST(%1\$s.meta_value AS DECIMAL(10,2)) BETWEEN '{$lo}' AND '{$hi}'" ) ), 'SELECT FOUND_ROWS()' );
+		},
+		'hot'    => function () use ( $price ) {
+			[ $lo, $hi ] = $price();
+			return array( flat_search( 'wpdo_hot_hp_listing', 'h', " AND `h`.`hp_price` BETWEEN {$lo} AND {$hi}" ), 'SELECT FOUND_ROWS()' );
+		},
+		'entity' => function () use ( $price ) {
+			[ $lo, $hi ] = $price();
+			return array( flat_search( 'wpdo_post_hp_listing_core', 'e', " AND `e`.`hp_price` BETWEEN '{$lo}' AND '{$hi}'" ), 'SELECT FOUND_ROWS()' );
+		},
+	),
+	'2. 價格區間 + verified=1'            => array(
+		'native' => function () use ( $price ) {
+			[ $lo, $hi ] = $price();
+			return array(
+				native_search(
+					array(
+						"%1\$s.meta_key = 'hp_price' AND CAST(%1\$s.meta_value AS DECIMAL(10,2)) BETWEEN '{$lo}' AND '{$hi}'",
+						"%1\$s.meta_key = 'hp_verified' AND %1\$s.meta_value = '1'",
+					)
+				),
+				'SELECT FOUND_ROWS()',
+			);
+		},
+		'hot'    => function () use ( $price ) {
+			[ $lo, $hi ] = $price();
+			return array( flat_search( 'wpdo_hot_hp_listing', 'h', " AND `h`.`hp_price` BETWEEN {$lo} AND {$hi} AND `h`.`hp_verified` = 1" ), 'SELECT FOUND_ROWS()' );
+		},
+		'entity' => function () use ( $price ) {
+			[ $lo, $hi ] = $price();
+			return array( flat_search( 'wpdo_post_hp_listing_core', 'e', " AND `e`.`hp_price` BETWEEN '{$lo}' AND '{$hi}' AND `e`.`hp_verified` = '1'" ), 'SELECT FOUND_ROWS()' );
+		},
+	),
+	'3. 某 vendor 名下 listing'          => array(
+		'post_parent' => fn() => array( 'SELECT SQL_CALC_FOUND_ROWS p.ID FROM `' . P . "posts` p WHERE p.post_parent = " . mt_rand( 1, $vendors ) . " AND p.post_type = 'hp_listing' AND ((p.post_status = 'publish')) ORDER BY p.post_date DESC LIMIT 0, 20", 'SELECT FOUND_ROWS()' ),
+		'native'      => fn() => array( native_search( array( "%1\$s.meta_key = 'hp_vendor' AND %1\$s.meta_value = '" . mt_rand( 1, $vendors ) . "'" ) ), 'SELECT FOUND_ROWS()' ),
+		'hot'         => fn() => array( flat_search( 'wpdo_hot_hp_listing', 'h', ' AND `h`.`hp_vendor` = ' . mt_rand( 1, $vendors ) ), 'SELECT FOUND_ROWS()' ),
+		'entity'      => fn() => array( flat_search( 'wpdo_post_hp_listing_core', 'e', " AND `e`.`hp_vendor` = '" . mt_rand( 1, $vendors ) . "'" ), 'SELECT FOUND_ROWS()' ),
+	),
+	'4. 單筆 listing 讀全部 meta'         => array(
+		'native' => fn() => array( 'SELECT post_id, meta_key, meta_value FROM `' . P . 'postmeta` WHERE post_id IN (' . random_ids( 1, $vendors, $listings ) . ') ORDER BY meta_id ASC' ),
+		'hot'    => fn() => array( 'SELECT * FROM `' . P . 'wpdo_hot_hp_listing` WHERE post_id = ' . random_ids( 1, $vendors, $listings ) ),
+		'entity' => fn() => array( 'SELECT * FROM `' . P . 'wpdo_post_hp_listing_core` WHERE post_id = ' . random_ids( 1, $vendors, $listings ) ),
+	),
+	'5. 列表頁預載 20 筆 meta'            => array(
+		'native' => fn() => array( 'SELECT post_id, meta_key, meta_value FROM `' . P . 'postmeta` WHERE post_id IN (' . random_ids( 20, $vendors, $listings ) . ') ORDER BY meta_id ASC' ),
+		'hot'    => fn() => array( 'SELECT * FROM `' . P . 'wpdo_hot_hp_listing` WHERE post_id IN (' . random_ids( 20, $vendors, $listings ) . ')' ),
+		'entity' => fn() => array( 'SELECT * FROM `' . P . 'wpdo_post_hp_listing_core` WHERE post_id IN (' . random_ids( 20, $vendors, $listings ) . ')' ),
+	),
+	'6. 單一 vendor 讀使用者 meta'        => array(
+		'native' => fn() => array( 'SELECT user_id, meta_key, meta_value FROM `' . P . 'usermeta` WHERE user_id IN (' . mt_rand( 1, $vendors ) . ') ORDER BY umeta_id ASC' ),
+		'entity' => fn() => array( 'SELECT * FROM `' . P . 'wpdo_user_hp_user` WHERE user_id = ' . mt_rand( 1, $vendors ) ),
+	),
+	'7. 寫入：更新一筆 listing 價格'      => array(
+		'native'     => fn() => array( 'UPDATE `' . P . "postmeta` SET meta_value = '" . mt_rand( 100, 100000 ) . ".00' WHERE post_id = " . random_ids( 1, $vendors, $listings ) . " AND meta_key = 'hp_price'" ),
+		'entity'     => fn() => array( 'UPDATE `' . P . "wpdo_post_hp_listing_core` SET hp_price = '" . mt_rand( 100, 100000 ) . "' WHERE post_id = " . random_ids( 1, $vendors, $listings ) ),
+		'dual_write' => function () use ( $vendors, $listings ) {
+			$id = random_ids( 1, $vendors, $listings );
+			$v  = mt_rand( 100, 100000 );
+			return array(
+				'UPDATE `' . P . "postmeta` SET meta_value = '{$v}.00' WHERE post_id = {$id} AND meta_key = 'hp_price'",
+				'UPDATE `' . P . "wpdo_post_hp_listing_core` SET hp_price = '{$v}' WHERE post_id = {$id}",
+			);
+		},
+	),
+);
+
+// ── Run ───────────────────────────────────────────────────────────────────
+
+/**
+ * @return array{ms: float, rows: int|string, timeout: bool}
+ */
+function run_statements( mysqli $db, array $statements ): array {
+	$is_write = str_starts_with( $statements[0], 'UPDATE' );
+	if ( $is_write ) {
+		$db->begin_transaction();
+	}
+	$rows    = 0;
+	$timeout = false;
+	$start   = hrtime( true );
+	foreach ( $statements as $n => $sql ) {
+		try {
+			$res = $db->query( $sql );
+		} catch ( mysqli_sql_exception $e ) {
+			if ( 1969 !== $e->getCode() ) { // ER_STATEMENT_TIMEOUT.
+				throw $e;
+			}
+			$timeout = true;
+			break;
+		}
+		if ( $res instanceof mysqli_result ) {
+			$rows = str_starts_with( $sql, 'SELECT FOUND_ROWS' ) ? (int) $res->fetch_row()[0] : ( 0 === $n ? $res->num_rows : $rows );
+			$res->free();
+		} elseif ( $is_write ) {
+			$rows = $db->affected_rows;
+		}
+	}
+	$ms = ( hrtime( true ) - $start ) / 1e6;
+	if ( $is_write ) {
+		$db->rollback();
+	}
+	return array( 'ms' => $ms, 'rows' => $rows, 'timeout' => $timeout );
+}
+
+function explain_keys( mysqli $db, string $sql ): string {
+	if ( ! str_starts_with( $sql, 'SELECT' ) ) {
+		return '—';
+	}
+	$keys = array();
+	$res  = $db->query( 'EXPLAIN ' . $sql );
+	while ( $row = $res->fetch_assoc() ) {
+		$keys[] = $row['table'] . ':' . ( $row['key'] ?? 'ALL' );
+	}
+	return implode( ', ', $keys );
+}
+
+function percentile( array $sorted, float $p ): float {
+	return $sorted[ (int) min( count( $sorted ) - 1, floor( $p * count( $sorted ) ) ) ];
+}
+
+printf( "Dataset: %d vendors, %d listings — %d iterations (+%d warm-up) per variant\n\n", $vendors, $listings, $iterations, $warmup );
+echo "| 情境 | 形式 | p50 ms | p95 ms | max ms | 結果筆數(中位) | 使用索引 |\n|---|---|---:|---:|---:|---:|---|\n";
+
+foreach ( $scenarios as $label => $variants ) {
+	foreach ( $variants as $variant => $build ) {
+		$times    = array();
+		$rows     = array();
+		$timeouts = 0;
+		$keys     = explain_keys( $db, $build()[0] );
+		for ( $i = 0; $i < $warmup + $iterations; $i++ ) {
+			$r = run_statements( $db, $build() );
+			if ( $i < $warmup ) {
+				continue;
+			}
+			$timeouts += (int) $r['timeout'];
+			$times[]   = $r['ms'];
+			$rows[]    = $r['rows'];
+		}
+		sort( $times );
+		sort( $rows );
+		printf(
+			"| %s | %s | %.2f | %.2f | %.2f | %s | %s |\n",
+			$label,
+			$variant . ( $timeouts ? " ⚠ {$timeouts}× timeout" : '' ),
+			percentile( $times, 0.5 ),
+			percentile( $times, 0.95 ),
+			end( $times ),
+			$rows[ intdiv( count( $rows ), 2 ) ],
+			$keys
+		);
+	}
+}