diff --git a/docs-2.0/1.introduction/3.nebula-graph-architecture/4.storage-service.md b/docs-2.0/1.introduction/3.nebula-graph-architecture/4.storage-service.md
index 0bff7d5769a..89cea25c141 100644
--- a/docs-2.0/1.introduction/3.nebula-graph-architecture/4.storage-service.md
+++ b/docs-2.0/1.introduction/3.nebula-graph-architecture/4.storage-service.md
@@ -142,6 +142,8 @@ EdgeA_Out and EdgeA_In are stored in storage layer with opposite directions, con
 
 Like EdgeA_Out and EdgeA_In, Nebula Graph redundantly stores the information of each edge, which doubles the actual capacities needed for edge storage. The key corresponding to the edge occupies a small hard disk space, but the space occupied by Value is proportional to the length and amount of the property value. Therefore, it will occupy a relatively large hard disk space if the property value of the edge is large or there are many edge property values.
 
+To ensure the final consistency of the two key-value pairs when operating on edges, enable the [TOSS function](../../5.configurations-and-logs/1.configurations/3.graph-config.md ). After that, the operation will be performed in Partition x first where the out-edge is located, and then in Partition y where the in-edge is located. Finally, the result is returned.
+
 ### Partition algorithm
 
 Nebula Graph uses a **static Hash** strategy to shard data through a modulo operation on vertex ID. All the out-keys, in-keys, and tag data will be placed in the same partition. In this way, query efficiency is increased dramatically.
diff --git a/docs-2.0/5.configurations-and-logs/1.configurations/3.graph-config.md b/docs-2.0/5.configurations-and-logs/1.configurations/3.graph-config.md
index 7f56fa5cec1..a56d4ebd7c7 100644
--- a/docs-2.0/5.configurations-and-logs/1.configurations/3.graph-config.md
+++ b/docs-2.0/5.configurations-and-logs/1.configurations/3.graph-config.md
@@ -92,3 +92,15 @@ For all parameters and their current values, see [Configurations](1.configuratio
 | ------------------- | ---------------- | ------------------------------------------ |
 |`enable_authorize`   |`false`           |When set to `false`, the system authentication is not enabled. For more information, see [Authentication](../../7.data-security/1.authentication/1.authentication.md).|
 |`auth_type`          |`password`        |Specifies the login method. Available values are `password`, `ldap`, and `cloud`.|
+
+## Experimental configurations
+
+| Name                | Predefined value         | Description                                |
+| ------------------- | ------------------------ | ------------------------------------------ |
+|`enable_experimental_feature`|`false`| Specifies the experimental feature. Optional values are `true` and `false`. For currently supported experimental features, see below.|
+
+### Experimental features
+
+| Name                 |  Description                             |
+| ------------------- | ------------------------------------------|
+| TOSS | The TOSS (Transaction on Storage Side) function is used to ensure the final consistency of the `INSERT`, `UPDATE`, or `UPSERT` operations on edges (because one edge logically corresponds to two key-value pairs on the hard disk). The `DELETE` operation is not currently supported. After the TOSS function is enabled, the time delay of related operations will be increased by about one time. |