数据库索引运行速度比预期慢

Question

我正在尝试运行此查询，在没有任何索引的情况下大约需要 76 秒，现在我在其中添加了索引，现在大约需要 80 秒。

该表目前有 24k 条记录。

这是 DDL :

CREATE TABLE personal_details (
    pno VARCHAR(20) PRIMARY KEY,   -- Personal number or unique ID
    first_name VARCHAR(20),        -- First name
    middle_name VARCHAR(20),       -- Middle name (nullable if not always used)
    last_name VARCHAR(20),         -- Last name
    dob DATE,                      -- Date of birth (YYYY-MM-DD format)
    gender CHAR(1),                -- Gender (M/F/O)
    email VARCHAR(50),             -- Email address
    phone_number VARCHAR(20),      -- Phone number (variable format, max 20 chars)
    address VARCHAR(255),          -- Address (full address string)
    city VARCHAR(50),              -- City
    state VARCHAR(50),             -- State
    zip_code VARCHAR(10),          -- Postal or zip code
    country VARCHAR(50),           -- Country
    marital_status VARCHAR(20),    -- Marital status (Single/Married/etc.)
    nationality VARCHAR(50),       -- Nationality
    occupation VARCHAR(50),        -- Job title or occupation
    salary DECIMAL(10, 2),         -- Salary (up to 10 digits, 2 decimal places)
    hire_date DATE,                -- Hire date (YYYY-MM-DD format)
    department VARCHAR(50),        -- Department name
    is_active BOOLEAN              -- Status flag for active/inactive (True/False)
);

这是需要 76 秒的查询：

SELECT
    pd1.pno AS pno1,
    pd1.first_name AS first_name1,
    pd1.last_name AS last_name1,
    pd2.pno AS pno2,
    pd2.first_name AS first_name2,
    pd2.last_name AS last_name2,
    pd3.pno AS pno3,
    pd3.first_name AS first_name3,
    pd3.last_name AS last_name3
FROM
    personal_details pd1
JOIN
    personal_details pd2 ON pd1.city = pd2.city
JOIN
    personal_details pd3 ON pd2.state = pd3.state
WHERE
    pd1.dob BETWEEN '1980-01-01' AND '1990-12-31'
    AND pd2.gender = 'M'
    AND pd3.salary > 50000
ORDER BY
    pd1.pno, pd2.pno, pd3.pno;

这是我创建的索引（尝试和测试了很少）

CREATE INDEX idx_pno ON personal_details (pno);

CREATE INDEX idx_composite_query ON personal_details (city, state, dob, gender, salary);

CREATE INDEX idx_city ON personal_details (city);

CREATE INDEX idx_state ON personal_details (state);

CREATE INDEX idx_dob ON personal_details (dob);

CREATE INDEX idx_gender ON personal_details (gender);

CREATE INDEX idx_salary ON personal_details (salary);

即使创建这些索引后，查询也花费了很多时间（大约 80 秒，比正常查询要长）

我试图了解索引是如何工作的，这就是为什么我创建了这个自连接的查询，我原以为执行时间会减少，但它增加了。

我尝试创建单个索引和复合索引，但都给了我几乎相同的结果。

尝试阅读几篇文章并根据其进行更改，但无济于事，有点迷失在这里。

我还尝试使用

EXPLAIN ANALYZE

分析查询

EXPLAIN
"-> Sort: pd1.pno, pd2.pno, pd3.pno  (actual time=19130..19461 rows=2.25e+6 loops=1)
    -> Stream results  (cost=346436 rows=105833) (actual time=129..1905 rows=2.25e+6 loops=1)
        -> Inner hash join (pd3.state = pd2.state)  (cost=346436 rows=105833) (actual time=129..634 rows=2.25e+6 loops=1)
            -> Filter: (pd3.salary > 50000)  (cost=50.4 rows=813) (actual time=0.0498..23.2 rows=22872 loops=1)
                -> Table scan on pd3  (cost=50.4 rows=24394) (actual time=0.0433..19.1 rows=24000 loops=1)
            -> Hash
                -> Nested loop inner join  (cost=7124 rows=3905) (actual time=0.105..62.5 rows=4916 loops=1)
                    -> Filter: ((pd2.gender = 'M') and (pd2.city is not null))  (cost=2633 rows=2439) (actual time=0.058..20.3 rows=7968 loops=1)
                        -> Table scan on pd2  (cost=2633 rows=24394) (actual time=0.0541..17.6 rows=24000 loops=1)
                    -> Index lookup on pd1 using idx_composite_query (city=pd2.city), with index condition: (pd1.dob between '1980-01-01' and '1990-12-31')  (cost=0.4 rows=1.6) (actual time=0.00432..0.00515 rows=0.617 loops=7968)
"

这就是向我展示的。此查询仅用于我自学的教育目的

Answer 1

您可以尝试添加这三个索引。其他索引可以去掉，没什么用

CREATE INDEX idx_bob_pno_state_city ON personal_details (bob, pno, state, city);

CREATE INDEX idx_city_gender_pno ON personal_details (city, gender, pno);

CREATE INDEX idx_state_salary_pno ON personal_details (city, salary, pno);