Advanced PostgreSQL: Indexing, Partitioning, Replication · 课时

管理分区表

学习添加、删除和修改分区的最佳实践，以及如何处理默认分区。

管理分区表 是 CoddyKit 上的免费 Advanced PostgreSQL: Indexing, Partitioning, Replication 课时。这是第 3 节课，共 4 节。你可以在下方免费阅读本课时的完整内容 — 然后在浏览器中使用内置代码编辑器和全天候 AI 导师进行实践。这是 Advanced PostgreSQL: Indexing, Partitioning, Replication 学习路径的一部分，你的进度在网页和 CoddyKit 应用中同步。 Advanced PostgreSQL: Indexing, Partitioning, Replication 课程共包含 4 节课。

本课时的部分内容尚未翻译，以英文显示。

Managing Partitioned Tables

As your database grows, managing partitioned tables becomes crucial. This lesson explores how to dynamically add, remove, and modify partitions to keep your data organized and performant.

We'll cover adding new partitions for future data, detaching old partitions for archiving, and attaching existing tables as new partitions. We'll also look at handling data that doesn't fit any defined partition.

Setting Up a Partitioned Table

Before we manage partitions, let's create a base partitioned table. We'll use a simple sales_data table, partitioned by sale_date using a range partitioning strategy.

This table will serve as our example throughout the lesson.

CREATE TABLE sales_data (
  id SERIAL,
  sale_date DATE,
  amount DECIMAL(10, 2)
) PARTITION BY RANGE (sale_date);

-- Add an initial partition for December 2023
CREATE TABLE sales_2023_12 PARTITION OF sales_data
FOR VALUES FROM ('2023-12-01') TO ('2024-01-01');

INSERT INTO sales_data (sale_date, amount) VALUES ('2023-12-15', 150.75);
SELECT * FROM sales_data;

Adding New Partitions

As time passes, you'll need to add new partitions for incoming data. This is typically done using the CREATE TABLE ... PARTITION OF ... command, similar to how we created the initial partition.

This method creates a new, empty child table and links it to the parent partitioned table for a specified range of values.

Example: Adding a New Month

Let's add a partition for January 2024. This ensures that any new sales data for January will automatically be directed to its own dedicated storage segment.

Notice how we define the specific date range for this new partition.

-- Add a partition for January 2024
CREATE TABLE sales_2024_01 PARTITION OF sales_data
FOR VALUES FROM ('2024-01-01') TO ('2024-02-01');

INSERT INTO sales_data (sale_date, amount) VALUES ('2024-01-20', 200.50);

-- Verify data is in the correct partition (optional check)
SELECT tableoid::regclass, * FROM sales_data WHERE sale_date = '2024-01-20';

Detaching Old Partitions

Over time, older partitions might contain historical data that is rarely accessed but still needs to be preserved. Detaching a partition allows you to remove it from the partitioned table structure without deleting the data.

The detached partition becomes a standalone, regular table, which can then be archived, moved, or even dropped if no longer needed.

Example: Archiving Old Data

Let's detach the sales_2023_12 partition. After detachment, it will exist as a separate table named sales_2023_12, no longer part of the sales_data partitioned table.

This is useful for reducing the active dataset size for queries.

-- Detach the December 2023 partition
ALTER TABLE sales_data DETACH PARTITION sales_2023_12;

-- The detached table still exists and contains its data
SELECT * FROM sales_2023_12;

-- Trying to query sales_data for Dec 2023 will now return nothing
SELECT * FROM sales_data WHERE sale_date = '2023-12-15';

Attaching Existing Tables

Sometimes you might have a large dataset in a regular table that you want to integrate into a partitioned table. Instead of inserting all data, you can attach that existing table as a new partition.

The table to be attached must be empty or contain only data that satisfies the partition's constraints. This is a very fast operation as it doesn't involve data movement.

Example: Attaching a Prepared Table

Imagine we have a standalone table, sales_2024_02_temp, containing February 2024 sales. We can quickly attach it to our sales_data partitioned table.

First, we create and populate the temporary table, then attach it.

-- Create a standalone table for February 2024 sales
CREATE TABLE sales_2024_02_temp (
  id SERIAL,
  sale_date DATE,
  amount DECIMAL(10, 2)
);
INSERT INTO sales_2024_02_temp (sale_date, amount) VALUES ('2024-02-10', 300.25);

-- Attach this table as a new partition
ALTER TABLE sales_data ATTACH PARTITION sales_2024_02_temp
FOR VALUES FROM ('2024-02-01') TO ('2024-03-01');

-- Now, sales_data can query the February data
SELECT * FROM sales_data WHERE sale_date = '2024-02-10';

Handling Undefined Data with DEFAULT

What happens if data comes in that doesn't fit any of your defined partition ranges? By default, it would cause an error. To prevent this, you can create a DEFAULT partition.

The default partition acts as a catch-all for any rows that don't match the criteria of other existing partitions. It's essential for robustness and preventing data loss.

Example: Creating a Default Partition

Let's add a default partition to our sales_data table. Then, we'll try inserting a sale from a future date (e.g., 2025) for which no explicit partition exists.

This data will automatically be routed to the sales_default partition.

-- Create a default partition
CREATE TABLE sales_default PARTITION OF sales_data DEFAULT;

-- Insert data that doesn't fit existing ranges (e.g., 2025)
INSERT INTO sales_data (sale_date, amount) VALUES ('2025-01-05', 450.99);

-- Verify data went to the default partition
SELECT tableoid::regclass, * FROM sales_data WHERE sale_date = '2025-01-05';

Partition Management Check

You've learned how to dynamically adjust your partitioned table structure.

Recap: Partition Management

In this lesson, you mastered key aspects of managing partitioned tables:

Adding new partitions: Use CREATE TABLE ... PARTITION OF ... to prepare for new data ranges.
Detaching partitions: The ALTER TABLE ... DETACH PARTITION ... command removes a partition from the set, turning it into an independent table for archiving or separate management.
Attaching partitions: Use ALTER TABLE ... ATTACH PARTITION ... to quickly integrate an existing regular table as a new partition.
Default partitions: Create a DEFAULT partition to catch any data that doesn't match other defined partition boundaries, ensuring data integrity.

These techniques empower you to maintain flexible, performant, and scalable database systems.

常见问题解答

「管理分区表」课时是免费的吗？

是的 — 「管理分区表」的完整文本可在网页上免费阅读。要进行交互式练习（内置代码编辑器和全天候 AI 导师）并解锁 Advanced PostgreSQL: Indexing, Partitioning, Replication 课程的其余内容，请升级到 CoddyKit PRO。 Advanced PostgreSQL: Indexing, Partitioning, Replication 课程共包含 4 节课。

「管理分区表」这节课中我会学到什么？

学习添加、删除和修改分区的最佳实践，以及如何处理默认分区。你通过在浏览器中直接运行的动手代码来练习 Advanced PostgreSQL: Indexing, Partitioning, Replication，全天候 AI 导师会在你学习这节课的过程中回答你的问题。

学习 Advanced PostgreSQL: Indexing, Partitioning, Replication 需要有经验吗？

无需任何先前经验。CoddyKit 上的 Advanced PostgreSQL: Indexing, Partitioning, Replication 课程适合初学者到高级学习者，你可以从这里开始或从头开始，按照自己的节奏学习。这是第 3 节课，共 4 节。

「管理分区表」课时需要多长时间？

大多数 CoddyKit 课程大约需要 5–10 分钟。每节课都很精短且互动，所以你能稳步进步，并在网页和应用中从离开的地方继续。

我能在这节 Advanced PostgreSQL: Indexing, Partitioning, Replication 课中编写并运行代码吗？

能。每节 Advanced PostgreSQL: Indexing, Partitioning, Replication 课都包含内置代码编辑器，你可以在浏览器中直接编写并运行真实代码，并获得即时 AI 反馈 — 无需本地设置。

此课程中的所有课时

← 返回 Advanced PostgreSQL: Indexing, Partitioning, Replication