This is an automated email from the ASF dual-hosted git repository.
zhangchen pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/doris-website.git
The following commit(s) were added to refs/heads/master by this push:
new 37c0e0316f [doc](delete) address comment delete overview (#1765)
37c0e0316f is described below
commit 37c0e0316fc15bc4412eaae5cdb748996b909054
Author: zhannngchen <zhangc...@selectdb.com>
AuthorDate: Tue Jan 14 14:06:45 2025 +0800
[doc](delete) address comment delete overview (#1765)
## Versions
- [x] dev
- [x] 3.0
- [x] 2.1
- [ ] 2.0
## Languages
- [x] Chinese
- [x] English
## Docs Checklist
- [x] Checked by AI
- [x] Test Cases Built
---
docs/data-operate/delete/delete-overview.md | 55 ++++++++++++----------
.../current/data-operate/delete/delete-overview.md | 49 ++++++++++---------
.../data-operate/delete/delete-overview.md | 49 ++++++++++---------
.../data-operate/delete/delete-overview.md | 49 ++++++++++---------
.../data-operate/delete/delete-overview.md | 55 ++++++++++++----------
.../data-operate/delete/delete-overview.md | 55 ++++++++++++----------
6 files changed, 177 insertions(+), 135 deletions(-)
diff --git a/docs/data-operate/delete/delete-overview.md
b/docs/data-operate/delete/delete-overview.md
index 52e7d43c78..bbb5e7b6ab 100644
--- a/docs/data-operate/delete/delete-overview.md
+++ b/docs/data-operate/delete/delete-overview.md
@@ -25,52 +25,59 @@ specific language governing permissions and limitations
under the License.
-->
-In Apache Doris, the delete operation is a key feature for managing and
cleaning data to meet the flexibility needs of users in large-scale data
analysis scenarios. Doris's deletion mechanism supports efficient logical
deletion and multi-version data management, achieving a good balance between
performance and flexibility.
+In Apache Doris, the delete operation is essential for managing and cleaning
data to meet the flexible needs of users in large-scale data analysis scenarios.
-## Implementation Mechanism of Deletion
+Doris offers a variety of delete functionalities, including DELETE statements,
delete sign, partition deletion, full table deletion, and atomic overwrite
using temporary partitions. The following sections will detail each feature:
-Doris's delete operation uses **logical deletion** rather than directly
physically deleting data. The core implementation mechanisms are as follows:
+### DELETE Statement
-1. **Logical Deletion**. The delete operation does not directly remove data
from storage but adds a delete marker to the target data. There are two main
ways to implement logical deletion: delete predicate and delete sign.
+The DELETE statement is the most commonly used method for deleting data and
supports all table models. Users can use it to delete data that meets specific
conditions.
- 1. Delete predicate is used for Duplicate and Aggregate models. Each
deletion directly records a conditional predicate on the corresponding dataset
to filter out the deleted data during queries.
- 2. Delete sign is used for the Unique Key model. Each deletion writes a
new batch of data to overwrite the data to be deleted, and the hidden column
`__DORIS_VERSION_COL__` of the new data is set to 1, indicating that the data
has been deleted.
- 3. Performance comparison: The operation speed of "delete predicate" is
very fast, whether deleting 1 row or 100 million rows, the speed is almost the
same, it just write a conditional predicate to the dataset; the write speed of
delete sign is proportional to the amount of data.
+The syntax of the DELETE statement is as follows:
-2. **Multi-Version Data Management**. Doris supports multi-version data (MVCC,
Multi-Version Concurrency Control), allowing concurrent operations on the same
dataset without affecting query results. The delete operation creates a new
version containing the delete marker, while the old version data is still
retained.
+```sql
+DELETE FROM table_name WHERE condition;
+```
-3. **Physical Deletion (Compaction)**. The periodically executed compaction
process cleans up data marked for deletion, thereby freeing up storage space.
This process is automatically completed by the system without user
intervention. Note that only Base Compaction will physically delete data, while
Cumulative Compaction only merges and reorders data, reducing the number of
rowsets and segments.
+While the DELETE statement can handle most deletion needs, it may not be the
most efficient in some scenarios. To address various deletion requirements
flexibly and efficiently, Doris also provides the following methods.
-## Use Cases for Delete Operations
+### Truncate Partition
-Doris provides various deletion methods to meet different needs:
+In Doris, managing data through date partitions and other methods is common.
Many users only need to retain data for a recent period (e.g., 7 days). For
expired data partitions, the truncate partition feature can be used for
efficient deletion.
-### Conditional Deletion
+Compared to the DELETE statement, truncate partition only requires modifying
some partition metadata to complete the deletion, making it the best method in
this scenario.
-Users can delete rows that meet specified conditions. For example:
+The syntax of partition deletion is as follows:
```sql
-DELETE FROM table_name WHERE condition;
+TRUNCATE TABLE tbl PARTITION(p1, p2);
```
-### Batch Deletion via data loading
-
-During data loading, logical deletion can be achieved by overwriting. This
method is suitable for batch deletion of a large number of keys or
synchronizing TP database deletions during CDC binlog synchronization.
+### Truncate Table
-### Deleting All Data
+Truncate table is suitable for quickly clearing a table while retaining its
structure, such as when redoing data in offline analysis scenarios.
-In some cases, data can be deleted by directly truncating the table or
partition. For example:
+The syntax of full truncate table is as follows:
```sql
TRUNCATE TABLE table_name;
```
-### Atomic Overwrite Using Temporary Partitions
+### Delete Sign
-In some cases, users may want to rewrite the data of a partition. If the data
is deleted and then imported, there will be a period when the data is
unavailable. In this case, users can create a corresponding temporary
partition, import the new data into the temporary partition, and then replace
the original partition atomically to achieve the goal.
+Data deletion can be considered a type of data update. Therefore, on the
primary key model (Unique Key) with update capabilities, users can use the
delete sign feature to perform delete operations as data updates.
+
+For example, in CDC data synchronization scenarios, the CDC program can mark a
DELETE operation binlog with a delete sign. When this data is written to Doris,
the corresponding primary key will be deleted.
+
+This method can perform batch deletion of a large number of primary keys,
which is more efficient than the DELETE statement.
+
+The delete sign is an advanced feature and is more complex to use compared to
the previous methods. For detailed usage, please refer to the document [Batch
Deletion](./delete-overview.md).
+
+### Atomic Overwrite Using Temporary Partitions
-## Notes
+In some cases, users want to rewrite the data of a partition. However, if the
data is deleted and then loaded, there will be a period when the data is
unavailable. In this case, users can first create a corresponding temporary
partition, load the new data into the temporary partition, and then replace the
original partition atomically. For detailed usage, please refer to the document
[Atomic Table Replacement](./atomicity-replace.md).
-1. The delete operation generates new data versions, so frequent deletions may
increase the number of versions, affecting query performance.
-2. Deleted data will still occupy storage until compaction is completed, so
the delete operation itself will not immediately reduce storage usage.
+## Precautions
+1. The delete operation will generate new data versions, so frequent deletions
may increase the number of versions, thereby affecting query performance.
+2. Deleted data will still occupy storage until merge and compression are
completed, so the delete operation itself will not immediately reduce storage
usage.
diff --git
a/i18n/zh-CN/docusaurus-plugin-content-docs/current/data-operate/delete/delete-overview.md
b/i18n/zh-CN/docusaurus-plugin-content-docs/current/data-operate/delete/delete-overview.md
index 715c7abc86..5737475290 100644
---
a/i18n/zh-CN/docusaurus-plugin-content-docs/current/data-operate/delete/delete-overview.md
+++
b/i18n/zh-CN/docusaurus-plugin-content-docs/current/data-operate/delete/delete-overview.md
@@ -24,52 +24,59 @@ specific language governing permissions and limitations
under the License.
-->
-在 Apache Doris 中，删除操作（Delete）是一项关键功能，用于管理和清理数据，以满足用户在大规模数据分析场景中的灵活性需求。Doris
的删除机制支持高效的标记删除和多版本数据管理，在性能和灵活性之间达到了良好的平衡。
+在 Apache Doris 中，删除操作（Delete）是一项关键功能，用于管理和清理数据，以满足用户在大规模数据分析场景中的灵活性需求。
-## 删除的实现机制
+Doris 提供了丰富多样的删除功能支持，包括：DELETE 语句、删除标记（delete
sign）、分区删除、全表删除以及使用临时分区实现原子覆盖写等功能。下面将详细介绍每一项功能：
-Doris 的删除操作采用**标记删除（Logical Deletion）** 的方式，而不是直接物理删除数据。以下是其核心实现机制：
+### DELETE 语句
-1. **标记删除**。删除操作不会直接从存储中移除数据，而是为目标数据添加一条删除标记。标记删除主要有两种实现方式：delete 谓词和 delete
sign。
+删除数据时最常用的是 DELETE 语句，该功能支持所有表模型，用户可以使用它删除符合条件的数据。
- 1. delete 谓词用于 Duplicate 模型和 Aggregate
模型，每次删除会直接在对应的数据集上记录一个条件谓词，用于在查询时过滤掉被删除的数据。
- 2. delete sign 用于 Unique Key 模型，每次删除会新写入一批数据覆盖要被删除的数据，同时新写入的数据会将隐藏列
`__DORIS_VERSION_COL__` 设置为 1，表示该数据已经被删除。
- 3. 性能比较：“delete 谓词”的操作速度非常快，无论是删除 1 条数据还是 1
亿条数据，速度都差不多——都是写一个条件谓词到数据集上；delete sign 的写入速度与数据量成正比。
+DELETE 语句的语法如下：
-2. **多版本数据管理**。Doris 支持多版本数据（MVCC，Multi-Version Concurrency
Control），允许在同一数据集上进行并发操作而不会影响查询结果。删除操作会创建一个新的版本，其中包含删除标记，而旧版本数据仍然被保留。
+```sql
+DELETE FROM table_name WHERE condition;
+```
-3.
**物理删除（Compaction）**。定期执行的合并压缩（Compaction）过程会清理标记为删除的数据，从而释放存储空间。此过程由系统自动完成，无需用户手动干预。注意，只有
Base Compaction 才会对数据进行物理删除，Cumulative Compaction 仅对数据进行合并及重新排序，减少 rowset 及
segment 数量。
+DELETE 语句基本能满足大部分用户在使用 Doris
过程中的删除需求，但在某些场景下它并不是最高效的。为了灵活高效地满足用户在各类场景的删除需求，Doris 还提供了如下几种删除方式。
-## 删除操作的使用场景
+### 分区删除
-Doris 提供多种删除方式，以满足不同场景的需求：
+在 Doris 中，通过日期分区等方式来管理数据是很常见的实践。很多用户只需要保留最近一段时间的数据（例如 7
天），对于过期的数据分区，可以采用分区删除（truncate partition）功能来进行高效的删除。
-### 条件删除
+相比 DELETE 语句，分区删除只需要修改一些分区元数据即可完成删除，是这种场景下最佳的删除方式。
-用户可以通过指定过滤条件，删除满足条件的行。例如：
+分区删除的语法如下：
```sql
-DELETE FROM table_name WHERE condition;
+TRUNCATE TABLE tbl PARTITION(p1, p2);
```
-### 通过导入进行批量删除
-
-在数据导入时，通过覆盖的方式实现逻辑删除。这种方式适用于批量删除大量的 key，或者在 CDC 同步 binlog 时同步 TP 数据库的删除操作。
+### 整表删除
-### 删除全部数据
+整表删除适用于快速清空表且保留表结构的场景，例如在离线分析场景中需要重做数据时。
-在某些情况下，可以通过直接清空表或分区实现对数据的删除，例如：
+整表删除的语法如下：
```sql
TRUNCATE TABLE table_name;
```
+### 删除标记（Delete Sign）
+
+数据删除可以视作数据更新的一种情况。因此，在具有更新能力的主键模型（Unique Key）上，用户可以通过删除标记功能，使用数据更新的方式实现删除操作。
+
+例如在 CDC 数据同步场景中，CDC 程序可以将一条 DELETE 操作的 binlog 打上删除标记，当这条数据写入 Doris
时，就会删除掉对应的主键。
+
+这种方式相对于 DELETE 语句来说，可以批量进行大量主键的删除操作，效率较高。
+
+删除标记属于高级功能，使用起来相比前几种要更复杂一些，详细的用法请参考文档[批量删除](./delete-overview.md)。
+
### 使用临时分区实现原子覆盖写
-某些情况下，用户希望能够重写某一分区的数据，但如果采用先删除再导入的方式进行，在中间会有一段时间无法查看数据。这时，用户可以先创建一个对应的临时分区，将新的数据导入到临时分区后，通过替换操作，原子性地替换原有分区，以达到目的。
+某些情况下，用户希望能够重写某一分区的数据，但如果采用先删除再导入的方式进行，在中间会有一段时间无法查看数据。这时，用户可以先创建一个对应的临时分区，将新的数据导入到临时分区后，通过替换操作，原子性地替换原有分区，以达到目的。详细用法请参考文档[表原子替换](./atomicity-replace.md)。
## 注意事项
1. 删除操作会生成新的数据版本，因此频繁执行删除可能会导致版本数量增加，从而影响查询性能。
2. 删除后的数据在合并压缩完成之前仍会占用存储，因此删除操作本身不会立即降低存储使用。
-
diff --git
a/i18n/zh-CN/docusaurus-plugin-content-docs/version-2.1/data-operate/delete/delete-overview.md
b/i18n/zh-CN/docusaurus-plugin-content-docs/version-2.1/data-operate/delete/delete-overview.md
index 715c7abc86..5737475290 100644
---
a/i18n/zh-CN/docusaurus-plugin-content-docs/version-2.1/data-operate/delete/delete-overview.md
+++
b/i18n/zh-CN/docusaurus-plugin-content-docs/version-2.1/data-operate/delete/delete-overview.md
@@ -24,52 +24,59 @@ specific language governing permissions and limitations
under the License.
-->
-在 Apache Doris 中，删除操作（Delete）是一项关键功能，用于管理和清理数据，以满足用户在大规模数据分析场景中的灵活性需求。Doris
的删除机制支持高效的标记删除和多版本数据管理，在性能和灵活性之间达到了良好的平衡。
+在 Apache Doris 中，删除操作（Delete）是一项关键功能，用于管理和清理数据，以满足用户在大规模数据分析场景中的灵活性需求。
-## 删除的实现机制
+Doris 提供了丰富多样的删除功能支持，包括：DELETE 语句、删除标记（delete
sign）、分区删除、全表删除以及使用临时分区实现原子覆盖写等功能。下面将详细介绍每一项功能：
-Doris 的删除操作采用**标记删除（Logical Deletion）** 的方式，而不是直接物理删除数据。以下是其核心实现机制：
+### DELETE 语句
-1. **标记删除**。删除操作不会直接从存储中移除数据，而是为目标数据添加一条删除标记。标记删除主要有两种实现方式：delete 谓词和 delete
sign。
+删除数据时最常用的是 DELETE 语句，该功能支持所有表模型，用户可以使用它删除符合条件的数据。
- 1. delete 谓词用于 Duplicate 模型和 Aggregate
模型，每次删除会直接在对应的数据集上记录一个条件谓词，用于在查询时过滤掉被删除的数据。
- 2. delete sign 用于 Unique Key 模型，每次删除会新写入一批数据覆盖要被删除的数据，同时新写入的数据会将隐藏列
`__DORIS_VERSION_COL__` 设置为 1，表示该数据已经被删除。
- 3. 性能比较：“delete 谓词”的操作速度非常快，无论是删除 1 条数据还是 1
亿条数据，速度都差不多——都是写一个条件谓词到数据集上；delete sign 的写入速度与数据量成正比。
+DELETE 语句的语法如下：
-2. **多版本数据管理**。Doris 支持多版本数据（MVCC，Multi-Version Concurrency
Control），允许在同一数据集上进行并发操作而不会影响查询结果。删除操作会创建一个新的版本，其中包含删除标记，而旧版本数据仍然被保留。
+```sql
+DELETE FROM table_name WHERE condition;
+```
-3.
**物理删除（Compaction）**。定期执行的合并压缩（Compaction）过程会清理标记为删除的数据，从而释放存储空间。此过程由系统自动完成，无需用户手动干预。注意，只有
Base Compaction 才会对数据进行物理删除，Cumulative Compaction 仅对数据进行合并及重新排序，减少 rowset 及
segment 数量。
+DELETE 语句基本能满足大部分用户在使用 Doris
过程中的删除需求，但在某些场景下它并不是最高效的。为了灵活高效地满足用户在各类场景的删除需求，Doris 还提供了如下几种删除方式。
-## 删除操作的使用场景
+### 分区删除
-Doris 提供多种删除方式，以满足不同场景的需求：
+在 Doris 中，通过日期分区等方式来管理数据是很常见的实践。很多用户只需要保留最近一段时间的数据（例如 7
天），对于过期的数据分区，可以采用分区删除（truncate partition）功能来进行高效的删除。
-### 条件删除
+相比 DELETE 语句，分区删除只需要修改一些分区元数据即可完成删除，是这种场景下最佳的删除方式。
-用户可以通过指定过滤条件，删除满足条件的行。例如：
+分区删除的语法如下：
```sql
-DELETE FROM table_name WHERE condition;
+TRUNCATE TABLE tbl PARTITION(p1, p2);
```
-### 通过导入进行批量删除
-
-在数据导入时，通过覆盖的方式实现逻辑删除。这种方式适用于批量删除大量的 key，或者在 CDC 同步 binlog 时同步 TP 数据库的删除操作。
+### 整表删除
-### 删除全部数据
+整表删除适用于快速清空表且保留表结构的场景，例如在离线分析场景中需要重做数据时。
-在某些情况下，可以通过直接清空表或分区实现对数据的删除，例如：
+整表删除的语法如下：
```sql
TRUNCATE TABLE table_name;
```
+### 删除标记（Delete Sign）
+
+数据删除可以视作数据更新的一种情况。因此，在具有更新能力的主键模型（Unique Key）上，用户可以通过删除标记功能，使用数据更新的方式实现删除操作。
+
+例如在 CDC 数据同步场景中，CDC 程序可以将一条 DELETE 操作的 binlog 打上删除标记，当这条数据写入 Doris
时，就会删除掉对应的主键。
+
+这种方式相对于 DELETE 语句来说，可以批量进行大量主键的删除操作，效率较高。
+
+删除标记属于高级功能，使用起来相比前几种要更复杂一些，详细的用法请参考文档[批量删除](./delete-overview.md)。
+
### 使用临时分区实现原子覆盖写
-某些情况下，用户希望能够重写某一分区的数据，但如果采用先删除再导入的方式进行，在中间会有一段时间无法查看数据。这时，用户可以先创建一个对应的临时分区，将新的数据导入到临时分区后，通过替换操作，原子性地替换原有分区，以达到目的。
+某些情况下，用户希望能够重写某一分区的数据，但如果采用先删除再导入的方式进行，在中间会有一段时间无法查看数据。这时，用户可以先创建一个对应的临时分区，将新的数据导入到临时分区后，通过替换操作，原子性地替换原有分区，以达到目的。详细用法请参考文档[表原子替换](./atomicity-replace.md)。
## 注意事项
1. 删除操作会生成新的数据版本，因此频繁执行删除可能会导致版本数量增加，从而影响查询性能。
2. 删除后的数据在合并压缩完成之前仍会占用存储，因此删除操作本身不会立即降低存储使用。
-
diff --git
a/i18n/zh-CN/docusaurus-plugin-content-docs/version-3.0/data-operate/delete/delete-overview.md
b/i18n/zh-CN/docusaurus-plugin-content-docs/version-3.0/data-operate/delete/delete-overview.md
index 715c7abc86..5737475290 100644
---
a/i18n/zh-CN/docusaurus-plugin-content-docs/version-3.0/data-operate/delete/delete-overview.md
+++
b/i18n/zh-CN/docusaurus-plugin-content-docs/version-3.0/data-operate/delete/delete-overview.md
@@ -24,52 +24,59 @@ specific language governing permissions and limitations
under the License.
-->
-在 Apache Doris 中，删除操作（Delete）是一项关键功能，用于管理和清理数据，以满足用户在大规模数据分析场景中的灵活性需求。Doris
的删除机制支持高效的标记删除和多版本数据管理，在性能和灵活性之间达到了良好的平衡。
+在 Apache Doris 中，删除操作（Delete）是一项关键功能，用于管理和清理数据，以满足用户在大规模数据分析场景中的灵活性需求。
-## 删除的实现机制
+Doris 提供了丰富多样的删除功能支持，包括：DELETE 语句、删除标记（delete
sign）、分区删除、全表删除以及使用临时分区实现原子覆盖写等功能。下面将详细介绍每一项功能：
-Doris 的删除操作采用**标记删除（Logical Deletion）** 的方式，而不是直接物理删除数据。以下是其核心实现机制：
+### DELETE 语句
-1. **标记删除**。删除操作不会直接从存储中移除数据，而是为目标数据添加一条删除标记。标记删除主要有两种实现方式：delete 谓词和 delete
sign。
+删除数据时最常用的是 DELETE 语句，该功能支持所有表模型，用户可以使用它删除符合条件的数据。
- 1. delete 谓词用于 Duplicate 模型和 Aggregate
模型，每次删除会直接在对应的数据集上记录一个条件谓词，用于在查询时过滤掉被删除的数据。
- 2. delete sign 用于 Unique Key 模型，每次删除会新写入一批数据覆盖要被删除的数据，同时新写入的数据会将隐藏列
`__DORIS_VERSION_COL__` 设置为 1，表示该数据已经被删除。
- 3. 性能比较：“delete 谓词”的操作速度非常快，无论是删除 1 条数据还是 1
亿条数据，速度都差不多——都是写一个条件谓词到数据集上；delete sign 的写入速度与数据量成正比。
+DELETE 语句的语法如下：
-2. **多版本数据管理**。Doris 支持多版本数据（MVCC，Multi-Version Concurrency
Control），允许在同一数据集上进行并发操作而不会影响查询结果。删除操作会创建一个新的版本，其中包含删除标记，而旧版本数据仍然被保留。
+```sql
+DELETE FROM table_name WHERE condition;
+```
-3.
**物理删除（Compaction）**。定期执行的合并压缩（Compaction）过程会清理标记为删除的数据，从而释放存储空间。此过程由系统自动完成，无需用户手动干预。注意，只有
Base Compaction 才会对数据进行物理删除，Cumulative Compaction 仅对数据进行合并及重新排序，减少 rowset 及
segment 数量。
+DELETE 语句基本能满足大部分用户在使用 Doris
过程中的删除需求，但在某些场景下它并不是最高效的。为了灵活高效地满足用户在各类场景的删除需求，Doris 还提供了如下几种删除方式。
-## 删除操作的使用场景
+### 分区删除
-Doris 提供多种删除方式，以满足不同场景的需求：
+在 Doris 中，通过日期分区等方式来管理数据是很常见的实践。很多用户只需要保留最近一段时间的数据（例如 7
天），对于过期的数据分区，可以采用分区删除（truncate partition）功能来进行高效的删除。
-### 条件删除
+相比 DELETE 语句，分区删除只需要修改一些分区元数据即可完成删除，是这种场景下最佳的删除方式。
-用户可以通过指定过滤条件，删除满足条件的行。例如：
+分区删除的语法如下：
```sql
-DELETE FROM table_name WHERE condition;
+TRUNCATE TABLE tbl PARTITION(p1, p2);
```
-### 通过导入进行批量删除
-
-在数据导入时，通过覆盖的方式实现逻辑删除。这种方式适用于批量删除大量的 key，或者在 CDC 同步 binlog 时同步 TP 数据库的删除操作。
+### 整表删除
-### 删除全部数据
+整表删除适用于快速清空表且保留表结构的场景，例如在离线分析场景中需要重做数据时。
-在某些情况下，可以通过直接清空表或分区实现对数据的删除，例如：
+整表删除的语法如下：
```sql
TRUNCATE TABLE table_name;
```
+### 删除标记（Delete Sign）
+
+数据删除可以视作数据更新的一种情况。因此，在具有更新能力的主键模型（Unique Key）上，用户可以通过删除标记功能，使用数据更新的方式实现删除操作。
+
+例如在 CDC 数据同步场景中，CDC 程序可以将一条 DELETE 操作的 binlog 打上删除标记，当这条数据写入 Doris
时，就会删除掉对应的主键。
+
+这种方式相对于 DELETE 语句来说，可以批量进行大量主键的删除操作，效率较高。
+
+删除标记属于高级功能，使用起来相比前几种要更复杂一些，详细的用法请参考文档[批量删除](./delete-overview.md)。
+
### 使用临时分区实现原子覆盖写
-某些情况下，用户希望能够重写某一分区的数据，但如果采用先删除再导入的方式进行，在中间会有一段时间无法查看数据。这时，用户可以先创建一个对应的临时分区，将新的数据导入到临时分区后，通过替换操作，原子性地替换原有分区，以达到目的。
+某些情况下，用户希望能够重写某一分区的数据，但如果采用先删除再导入的方式进行，在中间会有一段时间无法查看数据。这时，用户可以先创建一个对应的临时分区，将新的数据导入到临时分区后，通过替换操作，原子性地替换原有分区，以达到目的。详细用法请参考文档[表原子替换](./atomicity-replace.md)。
## 注意事项
1. 删除操作会生成新的数据版本，因此频繁执行删除可能会导致版本数量增加，从而影响查询性能。
2. 删除后的数据在合并压缩完成之前仍会占用存储，因此删除操作本身不会立即降低存储使用。
-
diff --git a/versioned_docs/version-2.1/data-operate/delete/delete-overview.md
b/versioned_docs/version-2.1/data-operate/delete/delete-overview.md
index 52e7d43c78..bbb5e7b6ab 100644
--- a/versioned_docs/version-2.1/data-operate/delete/delete-overview.md
+++ b/versioned_docs/version-2.1/data-operate/delete/delete-overview.md
@@ -25,52 +25,59 @@ specific language governing permissions and limitations
under the License.
-->
-In Apache Doris, the delete operation is a key feature for managing and
cleaning data to meet the flexibility needs of users in large-scale data
analysis scenarios. Doris's deletion mechanism supports efficient logical
deletion and multi-version data management, achieving a good balance between
performance and flexibility.
+In Apache Doris, the delete operation is essential for managing and cleaning
data to meet the flexible needs of users in large-scale data analysis scenarios.
-## Implementation Mechanism of Deletion
+Doris offers a variety of delete functionalities, including DELETE statements,
delete sign, partition deletion, full table deletion, and atomic overwrite
using temporary partitions. The following sections will detail each feature:
-Doris's delete operation uses **logical deletion** rather than directly
physically deleting data. The core implementation mechanisms are as follows:
+### DELETE Statement
-1. **Logical Deletion**. The delete operation does not directly remove data
from storage but adds a delete marker to the target data. There are two main
ways to implement logical deletion: delete predicate and delete sign.
+The DELETE statement is the most commonly used method for deleting data and
supports all table models. Users can use it to delete data that meets specific
conditions.
- 1. Delete predicate is used for Duplicate and Aggregate models. Each
deletion directly records a conditional predicate on the corresponding dataset
to filter out the deleted data during queries.
- 2. Delete sign is used for the Unique Key model. Each deletion writes a
new batch of data to overwrite the data to be deleted, and the hidden column
`__DORIS_VERSION_COL__` of the new data is set to 1, indicating that the data
has been deleted.
- 3. Performance comparison: The operation speed of "delete predicate" is
very fast, whether deleting 1 row or 100 million rows, the speed is almost the
same, it just write a conditional predicate to the dataset; the write speed of
delete sign is proportional to the amount of data.
+The syntax of the DELETE statement is as follows:
-2. **Multi-Version Data Management**. Doris supports multi-version data (MVCC,
Multi-Version Concurrency Control), allowing concurrent operations on the same
dataset without affecting query results. The delete operation creates a new
version containing the delete marker, while the old version data is still
retained.
+```sql
+DELETE FROM table_name WHERE condition;
+```
-3. **Physical Deletion (Compaction)**. The periodically executed compaction
process cleans up data marked for deletion, thereby freeing up storage space.
This process is automatically completed by the system without user
intervention. Note that only Base Compaction will physically delete data, while
Cumulative Compaction only merges and reorders data, reducing the number of
rowsets and segments.
+While the DELETE statement can handle most deletion needs, it may not be the
most efficient in some scenarios. To address various deletion requirements
flexibly and efficiently, Doris also provides the following methods.
-## Use Cases for Delete Operations
+### Truncate Partition
-Doris provides various deletion methods to meet different needs:
+In Doris, managing data through date partitions and other methods is common.
Many users only need to retain data for a recent period (e.g., 7 days). For
expired data partitions, the truncate partition feature can be used for
efficient deletion.
-### Conditional Deletion
+Compared to the DELETE statement, truncate partition only requires modifying
some partition metadata to complete the deletion, making it the best method in
this scenario.
-Users can delete rows that meet specified conditions. For example:
+The syntax of partition deletion is as follows:
```sql
-DELETE FROM table_name WHERE condition;
+TRUNCATE TABLE tbl PARTITION(p1, p2);
```
-### Batch Deletion via data loading
-
-During data loading, logical deletion can be achieved by overwriting. This
method is suitable for batch deletion of a large number of keys or
synchronizing TP database deletions during CDC binlog synchronization.
+### Truncate Table
-### Deleting All Data
+Truncate table is suitable for quickly clearing a table while retaining its
structure, such as when redoing data in offline analysis scenarios.
-In some cases, data can be deleted by directly truncating the table or
partition. For example:
+The syntax of full truncate table is as follows:
```sql
TRUNCATE TABLE table_name;
```
-### Atomic Overwrite Using Temporary Partitions
+### Delete Sign
-In some cases, users may want to rewrite the data of a partition. If the data
is deleted and then imported, there will be a period when the data is
unavailable. In this case, users can create a corresponding temporary
partition, import the new data into the temporary partition, and then replace
the original partition atomically to achieve the goal.
+Data deletion can be considered a type of data update. Therefore, on the
primary key model (Unique Key) with update capabilities, users can use the
delete sign feature to perform delete operations as data updates.
+
+For example, in CDC data synchronization scenarios, the CDC program can mark a
DELETE operation binlog with a delete sign. When this data is written to Doris,
the corresponding primary key will be deleted.
+
+This method can perform batch deletion of a large number of primary keys,
which is more efficient than the DELETE statement.
+
+The delete sign is an advanced feature and is more complex to use compared to
the previous methods. For detailed usage, please refer to the document [Batch
Deletion](./delete-overview.md).
+
+### Atomic Overwrite Using Temporary Partitions
-## Notes
+In some cases, users want to rewrite the data of a partition. However, if the
data is deleted and then loaded, there will be a period when the data is
unavailable. In this case, users can first create a corresponding temporary
partition, load the new data into the temporary partition, and then replace the
original partition atomically. For detailed usage, please refer to the document
[Atomic Table Replacement](./atomicity-replace.md).
-1. The delete operation generates new data versions, so frequent deletions may
increase the number of versions, affecting query performance.
-2. Deleted data will still occupy storage until compaction is completed, so
the delete operation itself will not immediately reduce storage usage.
+## Precautions
+1. The delete operation will generate new data versions, so frequent deletions
may increase the number of versions, thereby affecting query performance.
+2. Deleted data will still occupy storage until merge and compression are
completed, so the delete operation itself will not immediately reduce storage
usage.
diff --git a/versioned_docs/version-3.0/data-operate/delete/delete-overview.md
b/versioned_docs/version-3.0/data-operate/delete/delete-overview.md
index 52e7d43c78..bbb5e7b6ab 100644
--- a/versioned_docs/version-3.0/data-operate/delete/delete-overview.md
+++ b/versioned_docs/version-3.0/data-operate/delete/delete-overview.md
@@ -25,52 +25,59 @@ specific language governing permissions and limitations
under the License.
-->
-In Apache Doris, the delete operation is a key feature for managing and
cleaning data to meet the flexibility needs of users in large-scale data
analysis scenarios. Doris's deletion mechanism supports efficient logical
deletion and multi-version data management, achieving a good balance between
performance and flexibility.
+In Apache Doris, the delete operation is essential for managing and cleaning
data to meet the flexible needs of users in large-scale data analysis scenarios.
-## Implementation Mechanism of Deletion
+Doris offers a variety of delete functionalities, including DELETE statements,
delete sign, partition deletion, full table deletion, and atomic overwrite
using temporary partitions. The following sections will detail each feature:
-Doris's delete operation uses **logical deletion** rather than directly
physically deleting data. The core implementation mechanisms are as follows:
+### DELETE Statement
-1. **Logical Deletion**. The delete operation does not directly remove data
from storage but adds a delete marker to the target data. There are two main
ways to implement logical deletion: delete predicate and delete sign.
+The DELETE statement is the most commonly used method for deleting data and
supports all table models. Users can use it to delete data that meets specific
conditions.
- 1. Delete predicate is used for Duplicate and Aggregate models. Each
deletion directly records a conditional predicate on the corresponding dataset
to filter out the deleted data during queries.
- 2. Delete sign is used for the Unique Key model. Each deletion writes a
new batch of data to overwrite the data to be deleted, and the hidden column
`__DORIS_VERSION_COL__` of the new data is set to 1, indicating that the data
has been deleted.
- 3. Performance comparison: The operation speed of "delete predicate" is
very fast, whether deleting 1 row or 100 million rows, the speed is almost the
same, it just write a conditional predicate to the dataset; the write speed of
delete sign is proportional to the amount of data.
+The syntax of the DELETE statement is as follows:
-2. **Multi-Version Data Management**. Doris supports multi-version data (MVCC,
Multi-Version Concurrency Control), allowing concurrent operations on the same
dataset without affecting query results. The delete operation creates a new
version containing the delete marker, while the old version data is still
retained.
+```sql
+DELETE FROM table_name WHERE condition;
+```
-3. **Physical Deletion (Compaction)**. The periodically executed compaction
process cleans up data marked for deletion, thereby freeing up storage space.
This process is automatically completed by the system without user
intervention. Note that only Base Compaction will physically delete data, while
Cumulative Compaction only merges and reorders data, reducing the number of
rowsets and segments.
+While the DELETE statement can handle most deletion needs, it may not be the
most efficient in some scenarios. To address various deletion requirements
flexibly and efficiently, Doris also provides the following methods.
-## Use Cases for Delete Operations
+### Truncate Partition
-Doris provides various deletion methods to meet different needs:
+In Doris, managing data through date partitions and other methods is common.
Many users only need to retain data for a recent period (e.g., 7 days). For
expired data partitions, the truncate partition feature can be used for
efficient deletion.
-### Conditional Deletion
+Compared to the DELETE statement, truncate partition only requires modifying
some partition metadata to complete the deletion, making it the best method in
this scenario.
-Users can delete rows that meet specified conditions. For example:
+The syntax of partition deletion is as follows:
```sql
-DELETE FROM table_name WHERE condition;
+TRUNCATE TABLE tbl PARTITION(p1, p2);
```
-### Batch Deletion via data loading
-
-During data loading, logical deletion can be achieved by overwriting. This
method is suitable for batch deletion of a large number of keys or
synchronizing TP database deletions during CDC binlog synchronization.
+### Truncate Table
-### Deleting All Data
+Truncate table is suitable for quickly clearing a table while retaining its
structure, such as when redoing data in offline analysis scenarios.
-In some cases, data can be deleted by directly truncating the table or
partition. For example:
+The syntax of full truncate table is as follows:
```sql
TRUNCATE TABLE table_name;
```
-### Atomic Overwrite Using Temporary Partitions
+### Delete Sign
-In some cases, users may want to rewrite the data of a partition. If the data
is deleted and then imported, there will be a period when the data is
unavailable. In this case, users can create a corresponding temporary
partition, import the new data into the temporary partition, and then replace
the original partition atomically to achieve the goal.
+Data deletion can be considered a type of data update. Therefore, on the
primary key model (Unique Key) with update capabilities, users can use the
delete sign feature to perform delete operations as data updates.
+
+For example, in CDC data synchronization scenarios, the CDC program can mark a
DELETE operation binlog with a delete sign. When this data is written to Doris,
the corresponding primary key will be deleted.
+
+This method can perform batch deletion of a large number of primary keys,
which is more efficient than the DELETE statement.
+
+The delete sign is an advanced feature and is more complex to use compared to
the previous methods. For detailed usage, please refer to the document [Batch
Deletion](./delete-overview.md).
+
+### Atomic Overwrite Using Temporary Partitions
-## Notes
+In some cases, users want to rewrite the data of a partition. However, if the
data is deleted and then loaded, there will be a period when the data is
unavailable. In this case, users can first create a corresponding temporary
partition, load the new data into the temporary partition, and then replace the
original partition atomically. For detailed usage, please refer to the document
[Atomic Table Replacement](./atomicity-replace.md).
-1. The delete operation generates new data versions, so frequent deletions may
increase the number of versions, affecting query performance.
-2. Deleted data will still occupy storage until compaction is completed, so
the delete operation itself will not immediately reduce storage usage.
+## Precautions
+1. The delete operation will generate new data versions, so frequent deletions
may increase the number of versions, thereby affecting query performance.
+2. Deleted data will still occupy storage until merge and compression are
completed, so the delete operation itself will not immediately reduce storage
usage.
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@doris.apache.org
For additional commands, e-mail: commits-h...@doris.apache.org
