MySQL 8.4进阶之多表JOIN、索引与查询优化

USE school;
CREATE TABLE classes (
  id   INT PRIMARY KEY,
  name VARCHAR(20) NOT NULL
) ENGINE=InnoDB;
-- 给学生表补上 class_id（已有表可用 ALTER 追加）
ALTER TABLE students ADD COLUMN class_id INT, ADD INDEX idx_class (class_id);
CREATE TABLE scores (
  id         INT AUTO_INCREMENT PRIMARY KEY,
  student_id INT NOT NULL,
  course     VARCHAR(20) NOT NULL,
  score      DECIMAL(5,2),
  UNIQUE KEY uk_stu_course (student_id, course),   -- 一人一科一条，兼作索引
  INDEX idx_course (course)
) ENGINE=InnoDB;
INSERT INTO classes VALUES (1,'高一(1)班'), (2,'高一(2)班'), (3,'高一(3)班');
UPDATE students SET class_id = 1 WHERE id IN (1,2);
UPDATE students SET class_id = 2 WHERE id IN (3,4);
INSERT INTO scores (student_id, course, score) VALUES
  (1,'数学',92),(1,'英语',78),(2,'数学',55),(3,'数学',88),(3,'英语',91);

-- 不带 ON，得到 4 × 3 = 12 行笛卡尔积（几乎永远不是你要的）
SELECT * FROM students, classes;

SELECT s.name, c.name AS 班级, s.score
FROM students s
INNER JOIN classes c ON s.class_id = c.id;

-- 列出所有班级，以及各班的学生（没人就显示 NULL）
SELECT c.name AS 班级, s.name AS 学生
FROM classes c
LEFT JOIN students s ON s.class_id = c.id
ORDER BY c.id;

SELECT s.name
FROM students s
LEFT JOIN scores sc ON sc.student_id = s.id
WHERE sc.id IS NULL;      -- 右表主键为 NULL ⇒ 没匹配上

-- A：条件写在 ON 里 —— 先过滤右表，再左连接。左表行全保留
SELECT s.name, sc.score
FROM students s
LEFT JOIN scores sc ON sc.student_id = s.id AND sc.course = '数学';
-- B：条件写在 WHERE 里 —— 连接完再过滤，NULL 行被干掉，LEFT JOIN 退化成 INNER JOIN
SELECT s.name, sc.score
FROM students s
LEFT JOIN scores sc ON sc.student_id = s.id
WHERE sc.course = '数学';

-- 找出和"张三"同班的同学
SELECT s2.name
FROM students s1
JOIN students s2 ON s1.class_id = s2.class_id
WHERE s1.name = '张三' AND s2.name <> '张三';

-- 每个班的平均分、最高分、人数（没学生的班级也列出）
SELECT c.name AS 班级,
       COUNT(s.id)      AS 人数,
       ROUND(AVG(s.score),1) AS 平均分,
       MAX(s.score)     AS 最高分
FROM classes c
LEFT JOIN students s ON s.class_id = c.id
GROUP BY c.id, c.name
ORDER BY 平均分 DESC;

SELECT s.name, s.score, t.avg_score
FROM students s
JOIN (
  SELECT class_id, AVG(score) AS avg_score
  FROM students GROUP BY class_id
) t ON s.class_id = t.class_id
WHERE s.score > t.avg_score;

-- 查有数学成绩的学生
SELECT name FROM students WHERE id IN (SELECT student_id FROM scores WHERE course='数学');
SELECT name FROM students s WHERE EXISTS (SELECT 1 FROM scores sc WHERE sc.student_id=s.id AND sc.course='数学');
SELECT DISTINCT s.name FROM students s JOIN scores sc ON sc.student_id=s.id WHERE sc.course='数学';

CREATE INDEX idx_name ON students(name);                 -- 普通索引
CREATE UNIQUE INDEX uk_email ON students(email);         -- 唯一索引
CREATE INDEX idx_class_score ON students(class_id, score); -- 复合索引
ALTER TABLE students DROP INDEX idx_name;                -- 删除
SHOW INDEX FROM students;                                -- 查看

WHERE class_id = 1                         -- ✅ 用索引
WHERE class_id = 1 AND score > 80          -- ✅ 用两列
WHERE class_id = 1 AND score > 80 AND name='张三' -- ✅ 用三列
WHERE score > 80                           -- ❌ 跳过最左列，索引失效
WHERE class_id = 1 AND name = '张三'        -- ⚠️ 只能用上 class_id，name 被 score 挡住

-- 查询模式
WHERE class_id = ? AND score > ? ORDER BY created_at LIMIT 20;
-- 最优索引
CREATE INDEX idx_cls_score_time ON students(class_id, score, created_at);

CREATE INDEX idx_class_score ON students(class_id, score);
SELECT class_id, score FROM students WHERE class_id = 1;   -- ✅ 覆盖索引

EXPLAIN SELECT s.name FROM students s JOIN scores sc ON sc.student_id = s.id;

EXPLAIN ANALYZE
SELECT s.name, sc.score FROM students s JOIN scores sc ON sc.student_id = s.id;

1. WHERE YEAR(created_at) = 2026              -- ❌ 列上套函数
   WHERE created_at >= '2026-01-01' AND created_at < '2027-01-01'  -- ✅ 改写成范围
2. WHERE phone = 13800138000                  -- ❌ 隐式类型转换：phone 是 VARCHAR 却传数字
   WHERE phone = '13800138000'                -- ✅ 类型一致（这坑极其隐蔽，且报错也不报）
3. WHERE name LIKE '%三'                      -- ❌ 前导通配符
   WHERE name LIKE '张%'                      -- ✅ 前缀匹配可用索引
4. WHERE age + 1 > 20                         -- ❌ 列参与运算
   WHERE age > 19                             -- ✅
5. WHERE a = 1 OR b = 2                       -- ⚠️ OR 条件容易全表扫，改 UNION ALL 或分别建索引
6. WHERE status != 1 / NOT IN / IS NULL       -- ⚠️ 优化器可能放弃索引（取决于选择性）
7. 不满足最左前缀                              -- 见上文
8. JOIN 列字符集/排序规则不一致                -- utf8mb4_general_ci 连 utf8mb4_0900_ai_ci 会失效

-- 被驱动表（右表）的关联列没索引 → 外层每取一行，内层就全表扫一次
-- 1万 × 1万 = 1亿次比较，灾难
CREATE INDEX idx_student ON scores(student_id);

-- 强制/禁用哈希连接（8.4 支持）
SELECT /*+ HASH_JOIN(t1, t2) */    * FROM t1 JOIN t2 ON t1.c1 = t2.c1;
SELECT /*+ NO_HASH_JOIN(t1, t2) */ * FROM t1 JOIN t2 ON t1.c1 = t2.c1;

SELECT * FROM students ORDER BY id LIMIT 100000, 20;   -- ❌ 要扫 100020 行再丢弃前 10 万

SELECT s.* FROM students s
JOIN (SELECT id FROM students ORDER BY id LIMIT 100000, 20) t USING (id);

SELECT * FROM students WHERE id > 100000 ORDER BY id LIMIT 20;  -- ✅ 直接定位

ALTER TABLE students ALTER INDEX idx_name INVISIBLE;   -- 对优化器隐藏，但仍维护
-- 观察一段时间没问题后再 DROP
ALTER TABLE students ALTER INDEX idx_name VISIBLE;     -- 秒级回滚

EXPLAIN SELECT /*+ SET_VAR(optimizer_switch='use_invisible_indexes=on') */ * FROM students WHERE name='张三';

ALTER TABLE students ADD INDEX idx_year ((YEAR(created_at)));  -- 注意双括号！
SELECT * FROM students WHERE YEAR(created_at) = 2026;          -- 现在能走索引了

CREATE INDEX idx_score_desc ON students(score DESC, id ASC);
SELECT * FROM students ORDER BY score DESC, id ASC LIMIT 10;  -- 直接走索引顺序

ANALYZE TABLE students UPDATE HISTOGRAM ON class_id WITH 10 BUCKETS;

SELECT * FROM sys.schema_unused_indexes     WHERE object_schema = 'school';  -- 从没被用过
SELECT * FROM sys.schema_redundant_indexes  WHERE table_schema = 'school';   -- 被其他索引覆盖的冗余索引