ML Flow until January 24

[MaKarenina]

Hi! When i was creating a new endpoint a have this alert 

CREATE A MODEL SERVING ENDPOINT TO SERVE YOUR MODEL BEHIND A REST API INTERFACE. YOU CAN STILL USE LEGACY ML FLOW MODEL SERVING UNTIL JANUARY 2024

I don't understand if my Legacy MLFlow Model Serving models are going to stop working from January 2024, or if they will stop having support but will continue working.

Can you help me to figure this out?

[Kaniz]

Hi @MaKarenina, The alert you received states that you can continue using Legacy MLflow Model Serving until January 2024.

However, there are a few important points to consider:

1. Support: After January 2024, Legacy MLflow Model Serving will no longer receive official support. This means that if you encounter any issues or need assistance, there won’t be any official channels to address them.

2. Functionality: While support ends, your existing Legacy MLflow Model Serving models will likely continue to work. However, there won’t be any updates or bug fixes. So, if your models are functioning well and you don’t need additional features, they should continue to serve predictions as usual.

3. Transition to Model Serving: To stay up-to-date and benefit from the latest features, it’s recommended to transition to the new Model Serving experience built on serverless compute. This new approach provides better scalability, improved performance, and full support backed by the Azure Databricks SLA.

Here’s what you can do:

• Migrate to Model Serving: You can create a Model Serving endpoint and flexibly transition your model-serving workflows without disabling Legacy MLflow Model Serving. Follow these steps:

  1. Navigate to Serving endpoints in your machine learning workspace.
  2. Create a serving endpoint with your model using the new Model Serving approach.
  3. Update your application to use the new URL provided by the serving endpoint and adapt to the new scoring format.
  4. Once your models are transitioned, you can disable Legacy MLflow Model Serving for specific models.

• Migrate Deployed Model Versions: If you have different versions of your models (e.g., staging and production), you can create separate model serving endpoints for each version. This ensures a smooth transition. Refer to the documentation for details on creating separate endpoints for staging and production ver...¹.

Remember that while Legacy MLflow Model Serving will no longer be officially supported, your existing models should continue to function until you decide to transition. Consider migrating to the new Model Serving experience to take advantage of the latest capabilities and support.

Feel free to explore the documentation for more detailed instructions on the migration process. If you have any further questions, don’t hesitate to ask! 😊