Azure/BQ MCP Optimiser

The Azure/BQ MCP Optimiser is a powerful, extensible MCP server designed to provide FinOps and performance insights for both Google BigQuery and Microsoft Azure SQL Server. It features a modular architecture that allows for easy expansion to other data sources and includes an AI-powered tool to translate natural language questions into SQL queries.

Key Features

Dual Data Source Support: Connect to either Google BigQuery or Azure SQL Server by simply changing an environment variable.
Live Performance Monitoring: Get real-time insights into your database's performance, including cost analysis for BigQuery and resource usage (CPU, DTU) for Azure SQL.
Expensive Query Analysis: Identify the most resource-intensive queries in your database to pinpoint optimization opportunities.
AI-Powered Querying: Translate natural language questions into SQL queries using Azure OpenAI.
Extensible Architecture: The modular design, built around a BaseDataSource class, makes it easy to add support for other databases in the future.
Robust and Production-Ready: With structured logging and comprehensive error handling, the server is built to be reliable and easy to maintain.

Getting Started

1. Prerequisites

Python 3.8+
An active Google Cloud Platform account (for BigQuery) or Microsoft Azure account (for Azure SQL).

2. Installation

Clone the repository and install the required dependencies:

git clone <your-repo-url> cd <your-repo-name> pip install -r requirements.txt

3. Configuration

The application is configured using an .env file. A sample .env file is included in the repository. Rename it from .env.sample to .env and fill in the required values.

a. Choose Your Data Source

First, set the DATA_SOURCE_TYPE variable to either "bigquery" or "azuresql".

DATA_SOURCE_TYPE="bigquery"

b. Configure BigQuery

If you are using BigQuery, you will need to provide the following:

GCP_PROJECT_ID: Your Google Cloud project ID.
GOOGLE_APPLICATION_CREDENTIALS: The path to your GCP service account key file.

c. Configure Azure SQL

If you are using Azure SQL, you will need to provide the following:

AZURE_SQL_SERVER, AZURE_SQL_DATABASE, AZURE_SQL_USERNAME, AZURE_SQL_PASSWORD: Your Azure SQL database connection details.
AZURE_SUBSCRIPTION_ID, AZURE_RESOURCE_GROUP: Your Azure subscription and resource group names.
AZURE_TENANT_ID, AZURE_CLIENT_ID, AZURE_CLIENT_SECRET: Your Azure service principal credentials.
AZURE_OPENAI_API_KEY, AZURE_OPENAI_ENDPOINT, AZURE_OPENAI_DEPLOYMENT_NAME: Your Azure OpenAI service details.

4. Verify Your Configuration

Before running the server, you can use the verify_env.py script to check that your environment is configured correctly:

python verify_env.py

5. Run the Server

Once your configuration is verified, you can start the MCP server:

python mcp_server.py

Available Tools

The following tools are available through the MCP server:

get_cost_summary: Get a summary of your database's performance and cost.
get_expensive_queries: Get a list of the most resource-intensive queries.
get_project_costs: Get a breakdown of costs by project (for BigQuery) or database size (for Azure SQL).
get_cost_trends: Get historical data on cost or resource usage.
analyze_query_cost: Get an estimated execution plan for a query.
get_cost_by_user: Get a breakdown of database usage by user.
natural_language_to_sql: Translate a natural language question into a SQL query (Azure SQL only).