# Overview

Prism enables users to connect to a data warehouses via `Connector` objects. Connector objects are passed into the `PrismProject`'s instantiation via the [`connectors`](https://docs.runprism.com/fundamentals/prismproject-api#overview) keyword argument, and tasks can access connectors via [`CurrentRun.conn`](https://docs.runprism.com/fundamentals/currentrun-api/currentrun.conn).

Note that ***connector objects are entirely optional**.* You can just as easily create a separate connection in each of your downstream tasks. However, we like using `Connector` instances for the following reasons:

1. `Connector` instances are thread-safe. It automatically handles creating connection and cursor objects whenever you want to run SQL queries in a multi-threaded project. 
2. `Connector` instances can be shared across projects, facilitating collaboration and reducing duplicative code.

## Executing SQL

The `Connector` class has one public method: `execute_sql`. As the name suggests, this method allows users to execute SQL and return the results as a Python object.

Here is the full method definition:

```python
Connector.execute_sql(
    sql: str,
    return_type: Optional[Literal["pandas"]],
) -> Union[pd.DataFrame, List[List[Any]]]:
```

This method takes two arguments:

* `sql`: SQL query to execute
* `return_type`: return type — either `pandas` or `None`. If `pandas`, then the data is converted to a Pandas DataFrame. If `None`, then the data is returned as a list of tuples or dictionaries.

These are current `Connector` classes:

* BigQueryConnector
* PostgresConnector
* RedshiftConnector
* SnowflakeConnector
* TrinoConnector
* PrestoConnector