# Task Configuration Example: Parallel Import

+ **_`1)` Configuration_** 
  + **`backend = "local"`**
  + **`incremental-import = true`**
  + Set **`meta-schema-name`** to specify the name for the database as the coordinating registry 
    + Default value is **`lightning_metadata`**
+ **_`2)` Spin up multiple `TiDB Lightning` instances from their nodes_** 
```
[lightning]
level = "info"
file = "tidb-lightning.log"
~!@meta-schema-name = "lightning_metadata"@!~

[tikv-importer]
~!@backend = "local"@!~
~!@incremental-import = true@!~
sorted-kv-dir = "/mnt/ssd/sorted-kv-dir"

[mydumper]
data-source-dir = "/data/my_database"

[tidb]
host = "172.16.31.1"
port = 4000
user = "root"
password = ""
status-port = 10080
pd-addr = "172.16.31.4:2379"
```
-------------------------------------------------------------
Besides the mentioned parameters on the slide, no additional configuration is required for parallel import using `TiDB Lightning`. 

When `TiDB Lightning` is started, it registers meta data in the downstream `TiDB` cluster and automatically detects whether there are other instances migrating data to the target cluster at the same time. If there is, it automatically enters the parallel import mode.

When using the physical import mode to import data in parallel (which you should), ensure that there are no primary key or unique index conflicts between data sources, and between the tables in the target `TiDB` cluster, and there are no data writes in the target table during import. Otherwise, `TiDB Lightning` will fail to guarantee the correctness of the imported data, and the target table will contain inconsistent indexes after the import is completed.