Phoenix to Arize AX Migration

Seamlessly migrate your data from Phoenix to Arize AX

If you are looking to migrate to Arize AX but want to preserve your data, this tool will help migrate your traces, evaluations, annotations, datasets, and experiments from Phoenix.

Migration Tool Repo

Setup

Step 1: Download the Migration Tool

git clone https://github.com/Arize-ai/phoenix-to-ax-migration
cd phoenix-to-ax-migration

Step 2: Install Dependencies

pip install -r requirements.txt

Step 3: Configure Environment

Create a .env file with your configuration:

# Phoenix (include PHOENIX_API_KEY if using Phoenix Cloud)
PHOENIX_ENDPOINT="https://app.phoenix.arize.com/s/your-space-name"
PHOENIX_API_KEY=your-phoenix-api-key 

# Arize
ARIZE_API_KEY=your-arize-api-key
ARIZE_SPACE_ID=your-arize-space-id

# Export directory where Phoenix data will be saved locally
PHOENIX_EXPORT_DIR="phoenix_export"

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

Supported Data Types

Data Type	Details
Datasets & Experiments	• All datasets and their corresponding experiments • Experiment evaluations not yet migrated
Project Traces	• Contains traces, evaluation, and human annotations

⚠️ Import Warning for Annotations ⚠️

• Wait for traces to be indexed: After importing traces, you must wait a few minutes for them to be loaded and indexed in Arize AX before sending annotation data. The import process will prompt you to verify traces are available before proceeding with annotations.

• 31-day window: Only annotations for traces from the past 31 days can be logged to Arize. If your traces are older than 31 days, their annotations will be skipped with a warning message.

Usage

Export from Phoenix

# Export everything
python export_all_projects.py --all

# Export specific types
python export_all_projects.py --de      # datasets and experiments
python export_all_projects.py --traces   # traces with evaluations and annotations

Import to Arize

# Import everything
python import_to_arize.py --all

# Import specific types
python import_to_arize.py --de      # datasets and experiments
python import_to_arize.py --traces  # traces with evaluations and annotations

Generated Files

After export and import, you’ll see this structure:

phoenix_export/ # All data lives here
├── datasets/
│   ├── datasets.json
│   ├── dataset_{id}_examples.json
│   └── dataset_{id}_experiments.json
└── projects/
    └── {project_name}/
        ├── project_metadata.json
        ├── traces.json
        ├── evaluations.json
        └── annotations.json

results/  # Overview of import and export jobs
├── dataset_experiment_export_results.json
├── dataset_experiment_import_results.json
├── trace_export_results.json
└── trace_import_results.json

Troubleshooting

Import issues with traces

The formatting of some span attributes may not be compatible between Arize and Phoenix. We tried to cover as many cases as possible, but there may be some missing ones. If you encounter errors:

1. Check the results/ folder to see what errors have occured
2. Fix them in the project’s traces.json file
3. Re-import the data to Arize

Import issues with annotations

• Wait for traces to be fully indexed in Arize AX before importing annotations
• Check that your traces are within the 31-day window
• Review the results/ folder for detailed error messages

For Large Projects and Datasets

• Be patient - large imports can take time
• Monitor the results/ folder for progress

🎉 Congratulations - Your Phoenix data is now available in Arize AX!

More Support

• Detailed documentation: Check the GitHub repository
• Report issues: GitHub Issues