home / mcp / parquet mcp server

Parquet MCP Server

This read-only MCP Server allows you to connect to Parquet data from Claude Desktop through CData JDBC Drivers. For full CRUD support, check out our MCP Server for Parquet (https://www.cdata.com/drivers/parquet/download/mcp).

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "cdatasoftware-parquet-mcp-server-by-cdata": {
      "command": "java",
      "args": [
        "-jar",
        "/PATH/TO/CDataMCP-jar-with-dependencies.jar",
        "/PATH/TO/parquet.prp"
      ]
    }
  }
}

You can run a read-only MCP Server that exposes live Parquet data through a simple MCP interface. This enables AI clients to ask natural language questions and receive data results without writing SQL. The server wraps the CData JDBC Driver for Parquet and serves data via MCP tools, making Parquet data queryable in real time.

How to use

You will connect to the Parquet MCP Server using an MCP client. The server exposes read-only access to Parquet data and supports built-in tools to read table information, list columns, and run queries. As you interact with the MCP client, you can ask questions like which opportunities are open by a certain account or which calendar events you have today. The workflow is designed so you provide natural language questions and the client uses the available tools to retrieve results from Parquet-backed data.

How to install

Prerequisites you need before installing: - Java Runtime Environment (JRE) or Java Development Kit (JDK) installed on your machine - Maven for building the MCP server - A Parquet source accessible via CData JDBC Driver for Parquet - A valid license or trial for the CData JDBC Driver for Parquet (see licensing steps below)

1) Clone the project and build the MCP server ``git clone https://github.com/cdatasoftware/parquet-mcp-server-by-cdata.git cd parquet-mcp-server-by-cdata mvn clean install`This creates the MCP server JAR with dependencies namedCDataMCP-jar-with-dependencies.jar`.

2) Install the CData JDBC Driver for Parquet - Download the CData JDBC Driver for Parquet from the CData site - Install the driver following the installer prompts - Locate the driver JAR (for example, in the installation’s lib folder)

3) License the CData JDBC Driver - In the installation directory, run the license command to license the driver ``java -jar cdata.jdbc.parquet.jar --license`- Enter your name, email, andTRIAL` (or your license key) when prompted.

4) Configure your Parquet connection - Run the connection utility to build a connection string ``java -jar cdata.jdbc.parquet.jar`` - Test the connection and copy the resulting JDBC URL for storage

5) Create the Parquet MCP property file (parquet.prp) - Use the following properties to expose your Parquet data through MCP ``Prefix=parquet ServerName=CDataParquet ServerVersion=1.0 DriverPath=PATH\\TO\\cdata.jdbc.parquet.jar DriverClass=cdata.jdbc.parquet.ParquetDriver JdbcUrl=jdbc:parquet:InitiateOAuth=GETANDREFRESH; Tables=`- Save this asparquet.prp` in a known location

6) Create Claude Desktop (or any MCP client) configuration - Add an MCP server entry that points to the server JAR and the parquet.prp file. You will run the server via a Java command as described in the next step.

Starting the MCP server

Run the MCP server on the same machine as the client. Use the complete runtime command shown below.

java -jar /PATH/TO/CDataMCP-jar-with-dependencies.jar /PATH/TO/parquet.prp

Available tools

parquet_get_tables

Retrieves a list of tables available in the data source. Output is CSV with column headers on the first line.

parquet_get_columns

Retrieves a list of columns for a specified table. Output is CSV with column headers on the first line.

parquet_run_query

Executes a SQL SELECT query against the Parquet data source and returns results.