All pages
Powered by GitBook
1 of 1

Loading...

Phoenix to Arize AX Migration

Seamlessly migrate your data from Phoenix to Arize AX

Overview

This migration tool helps you move your data from Phoenix to Arize AX in just a few simple steps. It transfers:

  • 📊 Projects and Traces - Your observability data

  • 📝 Annotations - Feedback data

  • 🎯 Datasets - Training and test data

  • 🤖 Prompts - Prompt templates

  • 📈 Evaluations - Performance metrics

The migration happens in two steps: first export from Phoenix, then import to Arize AX.

What You Need

  • Access to your Phoenix instance

  • An Arize AX Pro account

  • Python installed on your computer

Getting Started

Step 1: Download the Migration Tool

git clone https://github.com/Dylancouzon/phoenix2ax
cd phoenix2ax
pip install -r requirements.txt

Note: This migration tool was created by an Arize team member for the community, but it is not an official Arize package.

Step 2: Configure Your Settings

  1. Copy the example configuration file:

    cp .env.example .env
  2. Edit the .env file with your details:

    # Your Phoenix instance
    PHOENIX_ENDPOINT=https://your-phoenix-instance.com
    PHOENIX_API_KEY=your-phoenix-api-key
    
    # Your Arize AX account  
    ARIZE_API_KEY=your-arize-api-key
    ARIZE_SPACE_ID=your-arize-space-id

You can find your Arize API key and Space ID in your Arize AX account settings.

Step 3: Export Your Data from Phoenix

Run this command to export all your data:

python export_all_projects.py --all

This creates a phoenix_export folder with all your data.

Step 4: Import Your Data to Arize AX

Run this command to import everything:

python import_to_arize.py --all

The import process will guide you through setting up annotations. Follow the prompts carefully.

Common Options

Export Specific Projects Only

If you only want certain projects:

python export_all_projects.py --all --project "my-important-project"

Export Specific Data Types

If you only need certain types of data:

# Just datasets and prompts
python export_all_projects.py --datasets --prompts

# Just traces
python export_all_projects.py --traces

Get More Details

Add --verbose to see detailed progress:

python export_all_projects.py --all --verbose
python import_to_arize.py --all --verbose

What Gets Created

After export, you'll see this structure:

phoenix_export/
├── datasets/          # Your datasets
├── prompts/           # Your prompt templates  
└── projects/          # Your projects and traces
    ├── project-1/
    └── project-2/

After import, check the results/ folder for detailed logs.

Troubleshooting

"Authentication failed"

  • Double-check your API keys in the .env file

  • Make sure your keys have the right permissions

"Connection timeout"

  • Check your internet connection

  • Try exporting one project at a time for large datasets

Import issues with annotations

  • Run the annotation setup first: python import_to_arize.py --annotations

  • Follow the setup guide in your Arize AX interface

  • Then re-run the full import

Getting more help

  • Use --verbose for detailed logs

  • Check the results/ folder for error details

  • Visit the GitHub repository for issues

Important Notes

For Annotations

Annotations require special setup:

  1. Run: python import_to_arize.py --annotations

  2. Configure annotation types in Arize AX (follow the prompts)

  3. Re-run the import after setup

For Large Datasets

If you have lots of data:

  • Export one project at a time

  • Use --verbose to monitor progress

  • Be patient - large imports can take time

Data Safety

  • Your original Phoenix data stays unchanged

  • You can re-run commands safely (duplicates are skipped)

  • All data transfers are encrypted

Next Steps

After migration:

  1. ✅ Check your data in Arize AX

  2. ✅ Verify key projects and datasets

  3. ✅ Test annotations and evaluations

  4. ✅ Set up your team's access in Arize AX


🎉 Congratulations! Your Phoenix data is now available in Arize AX with all relationships preserved.

Need More Help?

  • Detailed documentation: Check the GitHub repository

  • Report issues: GitHub Issues

  • Arize support: Arize Documentation