diff --git a/java-chassis-reference/en_US/assets/producer-default-executor.png b/java-chassis-reference/en_US/assets/producer-default-executor.png
new file mode 100644
index 0000000..0225f01
--- /dev/null
+++ b/java-chassis-reference/en_US/assets/producer-default-executor.png
Binary files differ
diff --git a/java-chassis-reference/en_US/build-provider/thread-pool.md b/java-chassis-reference/en_US/build-provider/thread-pool.md
new file mode 100644
index 0000000..2bf4d50
--- /dev/null
+++ b/java-chassis-reference/en_US/build-provider/thread-pool.md
@@ -0,0 +1,48 @@
+# Thread pool
+
+## Concept Description
+Thread pool is for executing synchronization business logic.  
+net send/receive or reactive business logic executing in eventloop, is independent of the thread pool  
+
+By default all synchronization methods are executed in a global built-in thread pool  
+If the business has special requirements, you can specify to use a custom global thread pool, and you can use separate thread pools according to the schemaId or operationId to achieve the effect of the isolated bin.  
+
+## Customize thread pool  
+* Implementing a thread pool  
+  Choose one of the following methods.  
+  * Implement the `java.util.concurrent.Executor` interface   
+    In order to support elegant exit, if the internal thread is not set to the daemon thread, you also need to implement the `java.io.Closeable` interface, responsible for destroying the thread pool.  
+  * Implement the `java.util.concurrent.ExecutorService` interface  
+* Declare the thread pool of the implementation as a spring bean  
+* Enable thread pool  
+  Suppose the new thread pool spring bean id is custom-executor  
+  * Replace the global thread pool  
+    servicecomb.executors.default: custom-executor
+  * Specify a thread pool dedicated to the schema    
+    servicecomb.executors.Provider.${schemaId}: custom-executor
+  * Specify a thread pool dedicated to the operation    
+    servicecomb.executors.Provider.${schemaId}.${operationId}: custom-executor
+  
+
+## ServiceComb built-in thread pool  
+In a general thread pool, all threads share a task queue. In this case, all network threads need to apply for the same queue to join the queue. All threads in the thread pool need to grab the task from the same queue. Throughput scenarios, which can lead to competitive conflicts and create performance bottlenecks  
+Therefore, in order to improve performance, ServiceComb's built-in thread pool is actually a wrapper of real thread pools, allowing multiple sets of thread pools to be configured inside, and each network thread is bound to a set of thread pools to reduce contention conflicts.  
+![](/assets/producer-default-executor.png)
+
+* Before version 1.2.0  
+
+| Configuration                                    | default      | Description                           |
+| :----------------------------------------------- | :----------- | :------------------------------------ |
+| servicecomb.executor.default.group               | 2            | Create several sets of thread pools   |
+| servicecomb.executor.default.thread-per-group    | CPU count    | Number of threads per thread pool     |
+
+* Version greater than or equal to 1.2.0
+
+| Configuration                                       | default           | Description                                                               |
+| :-------------------------------------------------- | :---------------- | :------------------------------------------------------------------------ |
+| servicecomb.executor.default.group                  | 2                 | Create several sets of thread pools                                       |
+| servicecomb.executor.default.thread-per-group       | 100               | Maximum number of threads per group of thread pools<br>Deprecated，new name：maxThreads-per-group        |
+| servicecomb.executor.default.coreThreads-per-group  | 25                | Minimum number of threads per group of thread pools<br>Threads are not pre-created, but after they have been created, only threads larger than this value will be destroyed by idle. |
+| servicecomb.executor.default.maxThreads-per-group   | 100               | Maximum number of threads per group of thread pools                       |
+| servicecomb.executor.default.maxIdleSecond-per-group| 60                | Each thread in the thread pool that exceeds coreThreads-per-group will destroy the thread if the idle timeout |
+| servicecomb.executor.default.maxQueueSize-per-group | Integer.MAX_VALUE | Maximum length of the task queue in each group of thread pools            |
diff --git a/java-chassis-reference/zh_CN/assets/producer-default-executor.png b/java-chassis-reference/zh_CN/assets/producer-default-executor.png
new file mode 100644
index 0000000..0225f01
--- /dev/null
+++ b/java-chassis-reference/zh_CN/assets/producer-default-executor.png
Binary files differ
diff --git a/java-chassis-reference/zh_CN/build-provider/thread-pool.md b/java-chassis-reference/zh_CN/build-provider/thread-pool.md
new file mode 100644
index 0000000..6d25efe
--- /dev/null
+++ b/java-chassis-reference/zh_CN/build-provider/thread-pool.md
@@ -0,0 +1,48 @@
+# 线程池
+
+## 说明
+线程池用于执行同步模式的业务逻辑。  
+网络收发及reactive模式的业务逻辑在Eventloop中执行，与线程池无关  
+
+默认所有同步方法都在一个全局内置线程池中执行  
+如果业务有特殊的需求，可以指定使用自定义的全局线程池，并且可以根据schemaId或operationId指定各自使用独立的线程池，实现隔离仓的效果  
+
+## 定制线程池
+* 实现线程池  
+  下面的方法任选其一即可
+  * 实现`java.util.concurrent.Executor`接口  
+    为了支持优雅退出，如果内部线程未设置为daemon线程，则还需要实现`java.io.Closeable`接口，负责销毁线程池
+  * 实现`java.util.concurrent.ExecutorService`接口
+* 将实现的线程池声明为spring bean
+* 启用线程池  
+  假设新线程池bean id为custom-executor
+  * 替换全局线程池  
+    servicecomb.executors.default: custom-executor
+  * 指定schema专用的线程池  
+    servicecomb.executors.Provider.${schemaId}: custom-executor
+  * 指定operation专用的线程池  
+    servicecomb.executors.Provider.${schemaId}.${operationId}: custom-executor
+  
+
+## ServiceComb内置线程池
+一般的线程池都是所有线程共享一个任务队列，在这种情况下，所有网络线程需要向同一个队列申请请求入队，线程池中的所有线程需要从同一个队列中抢任务执行，对于高吞吐的场景，这会导致竞争冲突，形成性能瓶颈  
+所以，为了提升性能，ServiceComb内置线程池实际是真正线程池的包装，允许在其内部配置多组线程池，且每个网络线程绑定一组线程池，以减小竞争冲突  
+![](/assets/producer-default-executor.png)
+
+* 1.2.0之前的版本
+
+| 配置项                                           | 默认值       | 含义                   |
+| :----------------------------------------------- | :----------- | :--------------------- |
+| servicecomb.executor.default.group               | 2            | 创建几组线程池         |
+| servicecomb.executor.default.thread-per-group    | CPU数        | 每组线程池的线程数     |
+
+* 大于等于1.2.0的版本
+
+| 配置项                                              | 默认值            | 含义                                                                      |
+| :-------------------------------------------------- | :---------------- | :------------------------------------------------------------------------ |
+| servicecomb.executor.default.group                  | 2                 | 创建几组线程池                                                            |
+| servicecomb.executor.default.thread-per-group       | 100               | 每组线程池的最大线程数<br>Deprecated，新名字：maxThreads-per-group        |
+| servicecomb.executor.default.coreThreads-per-group  | 25                | 每组线程池的最小线程数<br>线程不会预创建，而是已经创建后，只有大于这个值的线程，才会因idle而销毁 |
+| servicecomb.executor.default.maxThreads-per-group   | 100               | 每组线程池的最大线程数                                                    |
+| servicecomb.executor.default.maxIdleSecond-per-group| 60                | 每组线程池中超过coreThreads-per-group的线程，如果idle超时，则会销毁该线程 |
+| servicecomb.executor.default.maxQueueSize-per-group | Integer.MAX_VALUE | 每组线程池中任务队列的最大长度                                            |