# Create separate Gremlin Java client objects for read and write endpoints
<a name="best-practices-gremlin-java-separate"></a>

You can increase performance by only performing writes on the writer endpoint and reading from one or more read-only endpoints.

```
Client readerClient = Cluster.build("https://reader-endpoint")
          ...
          .connect()

Client writerClient = Cluster.build("https://writer-endpoint")
          ...
          .connect()
```