SQL复杂查询与性能优化：医药行业ERP系统实战指南

一、医药行业数据库特性分析

在医药ERP系统中，数据库通常包含以下核心表结构：

sql 复制代码

-- 药品主数据
CREATE TABLE drug_master (
    drug_id INT PRIMARY KEY,
    drug_name VARCHAR(255),
    specification VARCHAR(100),
    batch_number VARCHAR(50),
    expiration_date DATE,
    storage_conditions TEXT,
    approval_number VARCHAR(50) UNIQUE
);

-- 库存管理
CREATE TABLE inventory (
    inventory_id INT PRIMARY KEY,
    drug_id INT,
    warehouse_id INT,
    quantity DECIMAL(10,2),
    last_updated TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
    FOREIGN KEY (drug_id) REFERENCES drug_master(drug_id)
);

-- 订单管理
CREATE TABLE orders (
    order_id INT PRIMARY KEY,
    customer_id INT,
    order_date DATE,
    total_amount DECIMAL(10,2),
    status ENUM('pending','shipped','delivered','returned')
);

-- 物流运输
CREATE TABLE logistics (
    logistics_id INT PRIMARY KEY,
    order_id INT,
    carrier VARCHAR(50),
    tracking_number VARCHAR(100),
    departure_date DATE,
    arrival_date DATE,
    FOREIGN KEY (order_id) REFERENCES orders(order_id)
);

医药数据具有以下特点：

数据敏感性：包含患者隐私、药品专利信息等敏感数据
严格时效性：药品有效期管理要求精确到分钟
复杂关联关系：药品批次与库存、订单、物流的多对多关系
合规性要求：需符合FDA 21 CFR Part 11、GSP等法规

二、复杂查询核心技能实战

1. SQL执行顺序深度解析

sql 复制代码

-- 典型医药查询示例
SELECT 
    d.drug_name,
    SUM(i.quantity) AS total_stock,
    COUNT(DISTINCT o.order_id) AS order_count
FROM drug_master d
INNER JOIN inventory i ON d.drug_id = i.drug_id
LEFT JOIN orders o ON d.drug_id = o.drug_id
WHERE 
    d.expiration_date > CURRENT_DATE
    AND i.warehouse_id IN (101, 102, 103)
GROUP BY d.drug_id
HAVING total_stock > 500
ORDER BY total_stock DESC
LIMIT 10;

执行顺序可视化：
FROM drug_master INNER JOIN inventory WHERE条件过滤 LEFT JOIN orders GROUP BY聚合 HAVING过滤 SELECT投影 DISTINCT去重 ORDER BY排序 LIMIT分页

2. 多表连接优化策略

2.1 左连接与右连接的选择

sql 复制代码

-- 查询所有已过期但仍有库存的药品
SELECT d.drug_name, i.quantity
FROM drug_master d
LEFT JOIN inventory i 
    ON d.drug_id = i.drug_id
    AND i.quantity > 0
WHERE d.expiration_date < CURRENT_DATE;

2.2 笛卡尔积的风险控制

sql 复制代码

-- 安全写法：显式指定连接条件
SELECT c.carrier, l.tracking_number
FROM logistics l
CROSS JOIN carriers c
WHERE l.carrier = c.carrier_code;

3. 子查询与CTE的医药应用

3.1 药品批次追溯

sql 复制代码

-- 子查询实现
SELECT d.drug_name, o.order_date
FROM drug_master d
WHERE d.batch_number IN (
    SELECT batch_number 
    FROM recall_notices 
    WHERE recall_date BETWEEN '2024-01-01' AND '2024-12-31'
);

-- CTE优化
WITH recalled_batches AS (
    SELECT batch_number
    FROM recall_notices
    WHERE recall_date BETWEEN '2024-01-01' AND '2024-12-31'
)
SELECT d.drug_name, o.order_date
FROM drug_master d
JOIN recalled_batches rb 
    ON d.batch_number = rb.batch_number
JOIN orders o 
    ON d.drug_id = o.drug_id;

4. 窗口函数高级应用

4.1 库存周转率分析

sql 复制代码

SELECT 
    drug_id,
    quantity,
    DATE_TRUNC('month', last_updated) AS month,
    SUM(quantity) OVER (PARTITION BY drug_id ORDER BY DATE_TRUNC('month', last_updated)) AS rolling_stock,
    AVG(quantity) OVER (PARTITION BY drug_id) AS avg_stock
FROM inventory;

4.2 药品效期预警

sql 复制代码

SELECT 
    drug_name,
    expiration_date,
    DENSE_RANK() OVER (ORDER BY expiration_date) AS expiry_rank,
    LAG(expiration_date) OVER (ORDER BY expiration_date) AS previous_expiry
FROM drug_master;

5. CASE语句动态分类

sql 复制代码

-- 药品风险等级评估
SELECT 
    drug_name,
    CASE 
        WHEN expiration_date < CURRENT_DATE + INTERVAL '3 months' THEN '高风险'
        WHEN expiration_date < CURRENT_DATE + INTERVAL '6 months' THEN '中风险'
        ELSE '低风险'
    END AS risk_level
FROM drug_master;

三、性能优化核心技巧

1. 执行计划深度解析

sql 复制代码

EXPLAIN ANALYZE
SELECT *
FROM drug_master
WHERE approval_number = '国药准字H20230001';

关键指标解读：

指标名称	医药场景解读
Index Scan	理想情况，例如通过approval_number索引快速定位药品信息
Full Table Scan	需避免，常见于无索引的模糊查询
Rows Removed by Filter	过滤效率，高值说明大量数据在内存中过滤，应优化索引策略
Parallelism	并行查询，适用于大规模药品数据分析（需硬件支持）

2. 索引优化策略

2.1 复合索引设计

sql 复制代码

-- 创建复合索引示例
CREATE INDEX idx_drug_warehouse ON inventory(drug_id, warehouse_id);

-- 优化查询
SELECT *
FROM inventory
WHERE drug_id = 123 AND warehouse_id = 101;

2.2 覆盖索引实践

sql 复制代码

-- 创建覆盖索引
CREATE INDEX idx_drug_coverage ON drug_master(drug_name, specification, expiration_date) INCLUDE (approval_number);

-- 查询优化
SELECT drug_name, specification, expiration_date
FROM drug_master
WHERE approval_number = '国药准字H20230001';

3. 避免全表扫描

sql 复制代码

-- 反例：函数导致索引失效
SELECT *
FROM inventory
WHERE DATE_FORMAT(last_updated, '%Y-%m') = '2024-03';

-- 优化方案
SELECT *
FROM inventory
WHERE last_updated >= '2024-03-01'
AND last_updated < '2024-04-01';

4. 分页优化

sql 复制代码

-- 低效分页
SELECT *
FROM orders
LIMIT 100000, 10;

-- 优化方案
SELECT *
FROM orders
WHERE order_id > 100000
ORDER BY order_id
LIMIT 10;

5. 数据传输优化

sql 复制代码

-- 反例：传输冗余字段
SELECT *
FROM logistics
WHERE carrier = '顺丰速运';

-- 优化方案
SELECT logistics_id, tracking_number, arrival_date
FROM logistics
WHERE carrier = '顺丰速运';

6. 分区表设计

sql 复制代码

-- 按月份分区示例
CREATE TABLE order_history (
    order_id INT,
    order_date DATE,
    amount DECIMAL(10,2)
) PARTITION BY RANGE (YEAR(order_date) * 100 + MONTH(order_date)) (
    PARTITION p202301 VALUES LESS THAN (202301),
    PARTITION p202302 VALUES LESS THAN (202302),
    ...
);

四、高级优化场景实战

1. JOIN顺序优化

sql 复制代码

-- 小表驱动大表
SELECT *
FROM drug_categories dc
STRAIGHT_JOIN inventory i 
    ON dc.category_id = i.category_id
WHERE dc.category_name = '抗肿瘤药';

2. 临时表与物化视图

sql 复制代码

-- 创建临时表缓存高频查询
CREATE TEMPORARY TABLE temp_high_value_orders AS
SELECT order_id, drug_id, amount
FROM orders
WHERE amount > 10000;

-- 创建物化视图
CREATE MATERIALIZED VIEW mv_monthly_sales AS
SELECT 
    DATE_TRUNC('month', order_date) AS month,
    SUM(amount) AS total_sales
FROM orders
GROUP BY month;

3. 统计信息更新

sql 复制代码

-- MySQL手动更新统计信息
ANALYZE TABLE drug_master;

-- PostgreSQL
VACUUM ANALYZE drug_master;

4. 锁竞争优化

sql 复制代码

-- 乐观锁实现库存扣减
UPDATE inventory
SET quantity = quantity - 10
WHERE drug_id = 123
  AND quantity >= 10;

五、工具与调试

1. 执行计划分析工具

工具名称	医药行业适用场景
EXPLAIN FORMAT=JSON	MySQL复杂查询分析
EXPLAIN ANALYZE	PostgreSQL实时执行计划分析
pg_stat_statements	PostgreSQL慢查询监控

2. 监控方案设计

Prometheus 抓取数据库指标 Grafana 展示慢查询趋势显示锁等待情况监控索引使用率

3. 基准测试

bash 复制代码

# sysbench测试命令示例
sysbench \
  --test=oltp_read_write \
  --mysql-db=pharmacy_erp \
  --mysql-user=benchmark \
  --mysql-password=secure \
  --table_size=1000000 \
  run

六、常见误区与解决方案

1. 索引误用场景

sql 复制代码

-- 反例：低选择性字段索引
CREATE INDEX idx_low_selectivity ON drug_master(status);

-- 优化建议：仅为高选择性字段创建索引

2. NULL值处理

sql 复制代码

-- 错误写法
SELECT *
FROM drug_master
WHERE approval_number = NULL;

-- 正确写法
SELECT *
FROM drug_master
WHERE approval_number IS NULL;

3. 子查询优化

sql 复制代码

-- 反例： correlated subquery
SELECT drug_name
FROM drug_master d
WHERE EXISTS (
    SELECT 1
    FROM inventory i
    WHERE i.drug_id = d.drug_id
    AND i.quantity < 10
);

-- 优化方案：JOIN替代
SELECT DISTINCT d.drug_name
FROM drug_master d
JOIN inventory i 
    ON d.drug_id = i.drug_id
WHERE i.quantity < 10;

七、行业最佳实践

1. 数据归档策略

sql 复制代码

-- 按年归档历史订单
CREATE TABLE orders_archive_2023 AS
SELECT *
FROM orders
WHERE order_date < '2024-01-01';

-- 删除历史数据
DELETE FROM orders
WHERE order_date < '2024-01-01';

2. 数据脱敏处理

sql 复制代码

-- 使用函数进行脱敏
SELECT 
    order_id,
    CONCAT(LEFT(customer_name, 1), '***', RIGHT(customer_name, 1)) AS masked_name,
    MD5(customer_email) AS hashed_email
FROM orders;

3. 事务隔离级别设置

sql 复制代码

-- 可重复读隔离级别（默认）
SET SESSION TRANSACTION ISOLATION LEVEL REPEATABLE READ;

-- 更高一致性要求
SET SESSION TRANSACTION ISOLATION LEVEL SERIALIZABLE;

八、性能优化路线图

阶段	优化目标	关键措施
诊断阶段	识别性能瓶颈	执行计划分析、慢查询日志监控
优化阶段	消除全表扫描	创建合适索引、优化查询语句
巩固阶段	建立监控体系	Prometheus+Grafana实时监控
维护阶段	持续性能优化	定期更新统计信息、分析索引使用情况

九、总结与展望

在医药ERP系统中，高效的SQL查询和数据库性能直接影响业务运营效率和合规性。通过掌握复杂查询技术、应用性能优化策略、合理使用工具，并结合行业特性进行针对性设计，能够构建稳定、高效的数据库系统。未来随着AI技术的发展，自动化查询优化工具将成为趋势，值得医药行业从业者持续关注。