WARNING: THIS SITE IS A MIRROR OF GITHUB.COM / IT CANNOT LOGIN OR REGISTER ACCOUNTS / THE CONTENTS ARE PROVIDED AS-IS / THIS SITE ASSUMES NO RESPONSIBILITY FOR ANY DISPLAYED CONTENT OR LINKS / IF YOU FOUND SOMETHING MAY NOT GOOD FOR EVERYONE, CONTACT ADMIN AT ilovescratch@foxmail.com
Skip to content
This repository was archived by the owner on Jul 24, 2025. It is now read-only.
This repository was archived by the owner on Jul 24, 2025. It is now read-only.

Loading models from s3 #225

@bredamatt

Description

@bredamatt

Is your feature request related to a problem? Please describe.
PVCs are not necessarily the fastest storage medium in Kubernetes. Currently, it seems ModelService supports pvc and hf only.

S3 object stores and compatible APIs like those provided by MinIO for example, provide direct disk access, even when used inside K8s, thereby offering better rw performance when compared with PVCs.
Consequently, it would be good to support S3 as storage URIs for the model artefacts.

Describe the solution approach you'd like
Support s3:// in the modelArtifacts.uri parameter.

Describe alternatives you've considered
There's an alternative, such as localPath for example, however, this is insecure.

Additional context
It could also be good to consider a modelArtifacts.storageUri parameter in addition so that models can be pulled from hf but eventually stored in a local store, such as s3 for example, i.e. via init container for example.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions