CODEWITHSUNDEEP
uritrackingmlflowmlops
Ali Tber
Ali Tber

Reputation: 43

Difference between tracking_uri and the backend store uri in MLFLOW

I am using Mlflow for my project hosting it in an EC2 instance. I was wondering in MlFlow what is the difference between the backend_store_uri we set when we launch the server and the trarcking_uri ?

Thanks,

Upvotes: 3

Views: 2231

Answers (1)

Alex Ott
Alex Ott

Reputation: 87259

tracking_uri is the URL of the MLflow server (remote, or built-in in Databricks) that will be used to log metadata & model (see doc). In your case, this will be the URL pointing to your EC2 instance that should be configured in programs that will log parameters into your server.

backend_store_uri - is used by MLflow server to configure where to store this data - on filesystem, in SQL-compatible database, etc. (see doc). If you use SQL database, then you also need to provide the --default-artifact-root option to point where to store generated artifacts (images, model files, etc.)

Upvotes: 1

Related Questions