T-SQL 高级查询

简介

T-SQL（Transact-SQL）是 SQL Server 的扩展 SQL 语言。高级查询涵盖窗口函数、CTE 递归、PIVOT 转换、MERGE 合并、动态 SQL 等。掌握这些技巧能大幅提升 SQL 编写效率和查询性能。

特点

• 1.窗口函数 — 排名、分组聚合、偏移
• 2.CTE 递归 — 层级数据查询
• 3.PIVOT — 行转列
• 4.MERGE — 数据同步

窗口函数

ROW_NUMBER / RANK / DENSE_RANK

-- 排名函数
SELECT
    Name,
    Department,
    Salary,
    -- 不同排名方式
    ROW_NUMBER() OVER (PARTITION BY Department ORDER BY Salary DESC) AS RowNum,
    RANK()       OVER (ORDER BY Salary DESC) AS RankVal,
    DENSE_RANK() OVER (ORDER BY Salary DESC) AS DenseRankVal
FROM Employees;

-- 分页（替代 OFFSET）
WITH RankedEmployees AS (
    SELECT *,
        ROW_NUMBER() OVER (ORDER BY HireDate DESC) AS RowNum
    FROM Employees
    WHERE Department = 'IT'
)
SELECT * FROM RankedEmployees
WHERE RowNum BETWEEN 11 AND 20;  -- 第2页，每页10条

SUM / AVG / COUNT 窗口聚合

-- 累计销售额
SELECT
    OrderDate,
    Amount,
    SUM(Amount) OVER (ORDER BY OrderDate) AS RunningTotal,
    SUM(Amount) OVER (PARTITION BY YEAR(OrderDate) ORDER BY OrderDate) AS YearRunningTotal
FROM Orders;

-- 部门工资占比
SELECT
    Name,
    Department,
    Salary,
    Salary * 100.0 / SUM(Salary) OVER (PARTITION BY Department) AS PercentOfDept
FROM Employees;

-- 移动平均（最近3天）
SELECT
    OrderDate,
    Amount,
    AVG(Amount) OVER (ORDER BY OrderDate ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) AS MovingAvg3
FROM Orders;

LAG / LEAD

-- 环比增长
SELECT
    Month,
    Revenue,
    LAG(Revenue, 1) OVER (ORDER BY Month) AS PrevMonth,
    Revenue - LAG(Revenue, 1) OVER (ORDER BY Month) AS Growth,
    CAST((Revenue - LAG(Revenue, 1) OVER (ORDER BY Month)) * 100.0
        / NULLIF(LAG(Revenue, 1) OVER (ORDER BY Month), 0) AS DECIMAL(5,2)) AS GrowthRate
FROM MonthlyRevenue;

-- FIRST_VALUE / LAST_VALUE
SELECT
    ProductId,
    SaleDate,
    Price,
    FIRST_VALUE(Price) OVER (PARTITION BY ProductId ORDER BY SaleDate) AS FirstPrice,
    LAST_VALUE(Price) OVER (PARTITION BY ProductId ORDER BY SaleDate
        ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING) AS LastPrice
FROM ProductPrices;

CTE 公用表表达式

基本 CTE

-- CTE 替代子查询，提高可读性
WITH DepartmentStats AS (
    SELECT
        Department,
        COUNT(*) AS EmpCount,
        AVG(Salary) AS AvgSalary,
        MAX(Salary) AS MaxSalary
    FROM Employees
    GROUP BY Department
),
HighSalaryDepts AS (
    SELECT * FROM DepartmentStats WHERE AvgSalary > 15000
)
SELECT
    d.Department,
    d.EmpCount,
    d.AvgSalary,
    e.Name AS TopEarner
FROM HighSalaryDepts d
CROSS APPLY (
    SELECT TOP 1 Name FROM Employees e
    WHERE e.Department = d.Department
    ORDER BY e.Salary DESC
) e;

递归 CTE — 层级查询

-- 组织架构树
WITH OrgChart AS (
    -- 锚点：顶级领导
    SELECT
        Id, Name, ManagerId, Title,
        0 AS Level,
        CAST(Name AS NVARCHAR(1000)) AS Path
    FROM Employees
    WHERE ManagerId IS NULL

    UNION ALL

    -- 递归：逐级展开
    SELECT
        e.Id, e.Name, e.ManagerId, e.Title,
        o.Level + 1,
        CAST(o.Path + ' > ' + e.Name AS NVARCHAR(1000))
    FROM Employees e
    INNER JOIN OrgChart o ON e.ManagerId = o.Id
)
SELECT * FROM OrgChart ORDER BY Path;

-- BOM 物料清单展开
WITH BOM AS (
    SELECT ProductId, ComponentId, Quantity, 0 AS Level
    FROM ProductComponents
    WHERE ProductId = 'P001'

    UNION ALL

    SELECT pc.ProductId, pc.ComponentId, pc.Quantity * b.Quantity, b.Level + 1
    FROM ProductComponents pc
    INNER JOIN BOM b ON pc.ProductId = b.ComponentId
)
SELECT * FROM BOM WHERE Level > 0;

PIVOT 行转列

-- 月份作为列
SELECT *
FROM (
    SELECT Department, MONTH(SaleDate) AS MonthNum, Amount
    FROM Sales
    WHERE YEAR(SaleDate) = 2026
) s
PIVOT (
    SUM(Amount) FOR MonthNum IN ([1],[2],[3],[4],[5],[6],[7],[8],[9],[10],[11],[12])
) p;

-- 动态 PIVOT（需要动态 SQL）
DECLARE @cols NVARCHAR(MAX);
DECLARE @query NVARCHAR(MAX);

SELECT @cols = STRING_AGG(QUOTENAME(Category), ',')
FROM (SELECT DISTINCT Category FROM Products) t;

SET @query = N'
SELECT *
FROM (
    SELECT Category, Month, Sales
    FROM MonthlySales
) s
PIVOT (SUM(Sales) FOR Category IN (' + @cols + ')) p';

EXEC sp_executesql @query;

MERGE 数据同步

-- MERGE — 同步目标表和源表
MERGE INTO Products AS target
USING (
    SELECT Id, Name, Price, Category FROM StagingProducts
) AS source
ON target.Id = source.Id

-- 匹配时更新
WHEN MATCHED AND (
    target.Name <> source.Name OR
    target.Price <> source.Price OR
    target.Category <> source.Category
) THEN
    UPDATE SET
        target.Name = source.Name,
        target.Price = source.Price,
        target.Category = source.Category,
        target.UpdatedAt = GETUTCDATE()

-- 不匹配时插入
WHEN NOT MATCHED THEN
    INSERT (Id, Name, Price, Category)
    VALUES (source.Id, source.Name, source.Price, source.Category)

-- 目标有但源没有时删除
WHEN NOT MATCHED BY SOURCE THEN
    DELETE;

临时表与表变量

-- 临时表（# 开头）
CREATE TABLE #TempOrders (
    OrderId INT,
    TotalAmount DECIMAL(18,2),
    OrderDate DATETIME
);

INSERT INTO #TempOrders
SELECT Id, TotalAmount, OrderDate FROM Orders WHERE Status = 'Completed';

SELECT * FROM #TempOrders WHERE TotalAmount > 1000;
DROP TABLE #TempOrders;

-- 全局临时表（## 开头，跨会话）
CREATE TABLE ##GlobalStats (StatName NVARCHAR(100), StatValue INT);

-- 表变量
DECLARE @UserOrders TABLE (
    UserId INT,
    OrderCount INT,
    TotalAmount DECIMAL(18,2)
);

INSERT INTO @UserOrders
SELECT UserId, COUNT(*), SUM(TotalAmount)
FROM Orders
GROUP BY UserId;

SELECT * FROM @UserOrders WHERE OrderCount > 5;

性能优化技巧

-- 1. 使用 EXISTS 替代 IN
-- 差
SELECT * FROM Orders WHERE UserId IN (SELECT Id FROM VIPUsers);
-- 好
SELECT * FROM Orders WHERE EXISTS (SELECT 1 FROM VIPUsers WHERE VIPUsers.Id = Orders.UserId);

-- 2. 避免在 WHERE 中使用函数
-- 差
SELECT * FROM Orders WHERE YEAR(CreatedAt) = 2026;
-- 好
SELECT * FROM Orders WHERE CreatedAt >= '2026-01-01' AND CreatedAt < '2027-01-01';

-- 3. 使用 SET NOCOUNT ON
SET NOCOUNT ON;  -- 减少 DONE_IN_PROC 消息

-- 4. 批量操作
-- 差：逐行操作
-- 好：批量 INSERT / UPDATE

-- 5. 查看执行计划
SET STATISTICS IO ON;
SET STATISTICS TIME ON;
-- 查看逻辑读和 CPU 时间

窗口函数深入

高级窗口函数应用

-- 1. 连续区间问题（Gaps and Islands）
-- 场景：找出每个用户的连续登录天数
WITH login_groups AS (
    SELECT
        user_id,
        login_date,
        DATEADD(DAY, -ROW_NUMBER() OVER (PARTITION BY user_id ORDER BY login_date), login_date) AS grp
    FROM user_logins
),
continuous_days AS (
    SELECT
        user_id,
        grp,
        MIN(login_date) AS start_date,
        MAX(login_date) AS end_date,
        COUNT(*) AS consecutive_days
    FROM login_groups
    GROUP BY user_id, grp
)
SELECT * FROM continuous_days WHERE consecutive_days >= 3;

-- 2. 删除重复数据（保留最新一条）
WITH ranked AS (
    SELECT
        id,
        user_id,
        email,
        updated_at,
        ROW_NUMBER() OVER (PARTITION BY user_id ORDER BY updated_at DESC) AS rn
    FROM users
)
DELETE FROM ranked WHERE rn > 1;

-- 3. 百分位计算
SELECT
    department,
    employee_name,
    salary,
    PERCENTILE_CONT(0.5) WITHIN GROUP (ORDER BY salary) OVER (PARTITION BY department) AS median_salary,
    PERCENT_RANK() OVER (PARTITION BY department ORDER BY salary) AS pct_rank
FROM employees;

-- 4. 前后行对比
SELECT
    product_id,
        sales_date,
        sales_amount,
        LAG(sales_amount, 1) OVER (PARTITION BY product_id ORDER BY sales_date) AS prev_day_sales,
        sales_amount - LAG(sales_amount, 1) OVER (PARTITION BY product_id ORDER BY sales_date) AS day_over_day,
        LEAD(sales_amount, 7) OVER (PARTITION BY product_id ORDER BY sales_date) AS next_week_sales
FROM daily_sales;

-- 5. NTile 分桶
-- 将用户按消费金额分为 4 个等级
SELECT
    user_id,
    total_amount,
    NTILE(4) OVER (ORDER BY total_amount DESC) AS quartile
FROM user_consumption;

-- quartile = 1 为最高消费群体
-- quartile = 4 为最低消费群体

存储过程与触发器

存储过程

-- 分页查询存储过程
CREATE PROCEDURE usp_GetOrdersPaged
    @UserId INT = NULL,
    @Status NVARCHAR(20) = NULL,
    @PageIndex INT = 1,
    @PageSize INT = 20,
    @TotalCount INT OUTPUT
AS
BEGIN
    SET NOCOUNT ON;

    -- 计算总数
    SELECT @TotalCount = COUNT(*)
    FROM Orders
    WHERE (@UserId IS NULL OR UserId = @UserId)
      AND (@Status IS NULL OR Status = @Status);

    -- 分页查询
    SELECT
        Id, OrderNo, UserId, TotalAmount, Status, CreatedAt
    FROM Orders
    WHERE (@UserId IS NULL OR UserId = @UserId)
      AND (@Status IS NULL OR Status = @Status)
    ORDER BY CreatedAt DESC
    OFFSET (@PageIndex - 1) * @PageSize ROWS
    FETCH NEXT @PageSize ROWS ONLY;
END;
GO

-- 调用
DECLARE @total INT;
EXEC usp_GetOrdersPaged @UserId = 1001, @PageIndex = 2, @PageSize = 20, @TotalCount = @total OUTPUT;
SELECT @total AS TotalCount;

-- 带错误处理的存储过程
CREATE PROCEDURE usp_TransferFunds
    @FromAccountId INT,
    @ToAccountId INT,
    @Amount DECIMAL(18,2)
AS
BEGIN
    SET NOCOUNT ON;
    SET XACT_ABORT ON;  -- 遇到错误自动回滚
    SET TRANSACTION ISOLATION LEVEL SERIALIZABLE;

    BEGIN TRY
        BEGIN TRANSACTION;

        -- 检查余额
        DECLARE @balance DECIMAL(18,2);
        SELECT @balance = Balance FROM Accounts WHERE Id = @FromAccountId;

        IF @balance < @Amount
        BEGIN
            RAISERROR('余额不足', 16, 1);
            RETURN -1;
        END

        -- 扣款
        UPDATE Accounts SET Balance = Balance - @Amount WHERE Id = @FromAccountId;

        -- 入账
        UPDATE Accounts SET Balance = Balance + @Amount WHERE Id = @ToAccountId;

        -- 记录转账流水
        INSERT INTO Transfers (FromAccount, ToAccount, Amount, CreatedAt)
        VALUES (@FromAccountId, @ToAccountId, @Amount, GETUTCDATE());

        COMMIT TRANSACTION;
        RETURN 0;
    END TRY
    BEGIN CATCH
        IF @@TRANCOUNT > 0
            ROLLBACK TRANSACTION;

        -- 记录错误日志
        INSERT INTO ErrorLog (ErrorNumber, ErrorSeverity, ErrorState, ErrorProcedure, ErrorLine, ErrorMessage, ErrorTime)
        VALUES (ERROR_NUMBER(), ERROR_SEVERITY(), ERROR_STATE(), ERROR_PROCEDURE(), ERROR_LINE(), ERROR_MESSAGE(), GETUTCDATE());

        RETURN -1;
    END CATCH;
END;
GO

触发器

-- 审计触发器
CREATE TABLE AuditLog (
    AuditId BIGINT IDENTITY(1,1) PRIMARY KEY,
    TableName NVARCHAR(100) NOT NULL,
    Operation NVARCHAR(10) NOT NULL,
    PrimaryKeyId NVARCHAR(50),
    OldData NVARCHAR(MAX),
    NewData NVARCHAR(MAX),
    ChangedBy NVARCHAR(100),
    ChangedAt DATETIME2 DEFAULT SYSUTCDATETIME()
);

CREATE TRIGGER trg_Orders_Audit
ON Orders
AFTER INSERT, UPDATE, DELETE
AS
BEGIN
    SET NOCOUNT ON;

    -- INSERT 操作
    INSERT INTO AuditLog (TableName, Operation, PrimaryKeyId, NewData, ChangedBy)
    SELECT 'Orders', 'INSERT', CAST(Id AS NVARCHAR), 
           (SELECT * FROM inserted i WHERE i.Id = ins.Id FOR JSON PATH),
           SYSTEM_USER
    FROM inserted ins;

    -- DELETE 操作
    INSERT INTO AuditLog (TableName, Operation, PrimaryKeyId, OldData, ChangedBy)
    SELECT 'Orders', 'DELETE', CAST(Id AS NVARCHAR),
           (SELECT * FROM deleted d WHERE d.Id = del.Id FOR JSON PATH),
           SYSTEM_USER
    FROM deleted del;

    -- UPDATE 操作
    INSERT INTO AuditLog (TableName, Operation, PrimaryKeyId, OldData, NewData, ChangedBy)
    SELECT 'Orders', 'UPDATE', CAST(i.Id AS NVARCHAR),
           (SELECT * FROM deleted d WHERE d.Id = i.Id FOR JSON PATH),
           (SELECT * FROM inserted i WHERE i.Id = i.Id FOR JSON PATH),
           SYSTEM_USER
    FROM inserted i;
END;
GO

-- 注意事项：
-- 1. 触发器是隐式执行的，难以调试
-- 2. 触发器中的操作会增加事务时间
-- 3. 批量操作时触发器会逐行触发（SQL Server 2005+ 支持批量）
-- 4. 生产环境建议用 CDC（Change Data Capture）替代触发器

TRY_CAST 与错误处理

安全类型转换

-- TRY_CAST / TRY_CONVERT 失败返回 NULL 而不报错
SELECT TRY_CAST('123' AS INT);         -- 123
SELECT TRY_CAST('abc' AS INT);         -- NULL（不报错）
SELECT TRY_CONVERT(DATE, '2026-04-14'); -- 2026-04-14
SELECT TRY_CONVERT(DATE, 'invalid');    -- NULL

-- 实际应用：清洗脏数据
SELECT
    id,
    phone,
    CASE
        WHEN TRY_CAST(phone AS BIGINT) IS NOT NULL THEN 'valid'
        ELSE 'invalid'
    END AS phone_status
FROM users;

-- TRY_PARSE（依赖文化设置）
SELECT TRY_PARSE('1,234.56' AS DECIMAL(10,2));              -- 1234.56（英文文化）
SELECT TRY_PARSE('1.234,56' AS DECIMAL(10,2) USING 'de-DE'); -- 1234.56（德文文化）

-- IIF 简化 CASE 表达式
SELECT
    id,
    IIF(status = 1, 'active', 'inactive') AS status_text,
    IIF(TotalAmount >= 1000, 'VIP', '普通') AS customer_level
FROM orders;

-- CHOOSE 函数
SELECT
    id,
    CHOOSE(status, '待支付', '已支付', '已发货', '已完成', '已取消') AS status_text
FROM orders;

-- CONCAT_NULL_YIELDS_NULL 设置
SET CONCAT_NULL_YIELDS_NULL OFF;
SELECT CONCAT('Hello', NULL, 'World');  -- HelloWorld
SET CONCAT_NULL_YIELDS_NULL ON;
SELECT CONCAT('Hello', NULL, 'World');  -- HelloNULLWorld

临时表 vs 表变量 vs CTE

选择策略

-- 临时表（#table）
-- 适合：大量数据（> 1000 行）、需要索引、需要统计信息
CREATE TABLE #TempOrders (
    Id INT PRIMARY KEY,
    UserId INT,
    Amount DECIMAL(18,2),
    CreatedAt DATETIME
);
CREATE INDEX idx_temp_user ON #TempOrders(UserId);

-- 表变量（@table）
-- 适合：少量数据（< 1000 行）、短生命周期、存储过程内部
DECLARE @UserStats TABLE (
    UserId INT PRIMARY KEY,
    OrderCount INT,
    TotalAmount DECIMAL(18,2)
);
-- 表变量没有统计信息，不适合大查询

-- CTE（Common Table Expression）
-- 适合：一次性使用、提高可读性、递归查询
;WITH OrderStats AS (
    SELECT UserId, COUNT(*) AS OrderCount, SUM(Amount) AS TotalAmount
    FROM Orders
    GROUP BY UserId
)
SELECT * FROM OrderStats WHERE OrderCount > 10;

-- 表值函数
CREATE FUNCTION dbo.fn_GetUserOrders(@UserId INT)
RETURNS @Result TABLE (
    OrderId INT,
    OrderNo NVARCHAR(32),
    Amount DECIMAL(18,2),
    OrderDate DATETIME
)
AS
BEGIN
    INSERT INTO @Result
    SELECT Id, OrderNo, TotalAmount, CreatedAt
    FROM Orders WHERE UserId = @UserId;

    RETURN;
END;
GO

-- 选择指南：
-- | 场景              | 推荐         | 原因                     |
-- |-------------------|-------------|-------------------------|
-- | < 1000 行         | 表变量       | 无日志开销，性能好         |
-- | > 1000 行         | 临时表       | 有统计信息，优化器可选择计划 |
-- | 需要索引          | 临时表       | 表变量不支持创建索引       |
-- | 递归              | CTE         | 语法简洁                 |
-- | 可复用            | 表值函数     | 可参数化，可复用          |
-- | 跨批处理使用      | 临时表       | 表变量在批处理结束时销毁   |

JSON 操作（SQL Server 2016+）

JSON 查询与修改

-- JSON 数据存储
CREATE TABLE products (
    id INT PRIMARY KEY,
    name NVARCHAR(200),
    attributes NVARCHAR(MAX)  -- 存储 JSON 字符串
);

INSERT INTO products VALUES
(1, 'Laptop', '{"brand":"Dell","cpu":"i7","ram":"16GB","storage":"512GB SSD"}'),
(2, 'Phone', '{"brand":"Apple","model":"iPhone 15","color":"black"}');

-- JSON_VALUE 提取标量值
SELECT
    id,
    name,
    JSON_VALUE(attributes, '$.brand') AS brand,
    JSON_VALUE(attributes, '$.cpu') AS cpu
FROM products;

-- JSON_QUERY 提取对象/数组
SELECT
    id,
    JSON_QUERY(attributes, '$.storage') AS storage_info
FROM products;

-- OPENJSON 将 JSON 转为表
SELECT *
FROM OPENJSON((SELECT attributes FROM products WHERE id = 1))
WITH (
    brand NVARCHAR(50) '$.brand',
    cpu NVARCHAR(50) '$.cpu',
    ram NVARCHAR(50) '$.ram'
);

-- JSON 修改（SQL Server 2017+）
UPDATE products
SET attributes = JSON_MODIFY(attributes, '$.ram', '32GB')
WHERE id = 1;

-- JSON_MODIFY 添加新属性
UPDATE products
SET attributes = JSON_MODIFY(
    JSON_MODIFY(attributes, '$.price', 6999),
    '$.discount', 0.95
)
WHERE id = 1;

-- JSON 索引（计算列 + 索引）
ALTER TABLE products
ADD brand AS JSON_VALUE(attributes, '$.brand');

CREATE INDEX idx_products_brand ON products(brand);

SELECT * FROM products WHERE brand = 'Dell';

优点

• 1.窗口函数 — 复杂分析查询变得简单
• 2.CTE 递归 — 优雅处理层级数据
• 3.MERGE — 一次操作完成同步
• 4.可读性 — CTE 和窗口函数提高 SQL 可读性

缺点

• 1.学习曲线 — 窗口函数语法复杂
• 2.性能陷阱 — 不当使用可能导致性能问题
• 3.调试困难 — 复杂 SQL 排错不易
• 4.方言限制 — 部分 T-SQL 语法不通用

总结

T-SQL 高级查询是后端开发者的核心竞争力。窗口函数处理排名和累计，CTE 递归处理层级数据，PIVOT 处理行列转换，MERGE 处理数据同步。核心原则：优先用集合操作替代游标，用 CTE 提高可读性，用窗口函数替代自连接。

关键知识点

• 数据库主题一定要同时看数据模型、读写模式和执行代价。
• 很多性能问题不是 SQL 语法问题，而是索引、统计信息、事务和数据分布问题。
• 高可用、备份、迁移和治理与查询优化同样重要。
• 先把数据模型、访问模式和执行代价绑定起来理解。

项目落地视角

• 所有优化前后都保留执行计划、样本 SQL 和关键指标对比。
• 上线前准备回滚脚本、备份点和校验方案。
• 把连接池、锁等待、慢查询和容量增长纳入日常巡检。
• 保留执行计划、样本 SQL、索引定义和优化前后指标。

常见误区

• 脱离真实数据分布讨论索引或分片。
• 只看单条 SQL，不看整条业务链路的事务和锁。
• 把测试环境结论直接等同于生产环境结论。
• 脱离真实数据分布设计索引。

进阶路线

• 继续向执行计划、存储引擎、复制机制和数据治理层深入。
• 把主题与 ORM、缓存、消息队列和归档策略联动起来思考。
• 沉淀成数据库设计规范、SQL 审核规则和变更流程。
• 继续深入存储引擎、复制机制、归档与冷热分层治理。

适用场景

• 当你准备把《T-SQL 高级查询》真正落到项目里时，最适合先在一个独立模块或最小样例里验证关键路径。
• 适合数据建模、查询优化、事务控制、高可用和迁移治理。
• 当系统开始遇到慢查询、锁冲突、热点数据或容量增长时，这类主题价值最高。

落地建议

• 先分析真实查询模式、数据量级和写入特征，再决定索引或分片策略。
• 所有优化结论都结合执行计划、样本数据和监控指标验证。
• 高风险操作前准备备份、回滚脚本与校验 SQL。

排错清单

• 先确认瓶颈在 CPU、I/O、锁等待、网络还是 SQL 本身。
• 检查执行计划是否走错索引、是否发生排序或全表扫描。
• 排查长事务、隐式类型转换、统计信息过期和参数嗅探。

复盘问题

• 如果把《T-SQL 高级查询》放进你的当前项目，最先要验证的输入、输出和失败路径分别是什么？
• 《T-SQL 高级查询》最容易在什么规模、什么边界条件下暴露问题？你会用什么指标或日志去确认？
• 相比默认实现或替代方案，采用《T-SQL 高级查询》最大的收益和代价分别是什么？

延伸阅读

• Redis 高级应用
• PostgreSQL advanced
• MongoDB 高级应用
• EF Core 高级查询