Namespaces in a ML Serving project are organization entities for your models. It allows user to simply classify their models depending on their needs by attaching them to an existing namespace.
Each namespace is linked to an object storage container of the same public cloud project where users can upload their serialized machine learning models. That object storage container is configurable.
- Number of created namespaces is restricted to
100 maximumper public cloud user.
- Each namespace is linked to one and only object storage container of the same public cloud project.
- Each namespace is linked to one and only docker registry.
- Each namespace can contain as many models as the user wants.
- Each model deployed inside a namespace will be reachable from the
same base url such as
<id-of-namespace>will be a generated unique identifier of the namespace and
<cx>the identifier of your assigned cluster.
Under the hood
- You can check the OVHcloud documention on how to create a namespace.