How Will You Use Image Volumes in Kubernetes v1.33?

Wouldn't you typically serve this data through a standard PV/PVC setup? File servers excel at delivering files, and I'm having a hard time spotting a strong use case for OCI volumes— I wouldn't want bulky data to bloat my source control.

Using OCI artifacts is a smart approach! It allows you to distribute various types of data. For example, in my homelab, I have a CI setup that pulls DNS records from Cloudflare and generates related files. Mounting the OCI volume simplifies updates and trades in the benefits of an OCI setup, like SBOM and versioning.

I play around with LLM workloads on my homelab. Managing model size can be tricky; right now, I sync models directly with NFS on my two GPU nodes. The idea of handling a 300+GB image without packing it all together sounds way more manageable!

Wow, decoupling blobs from business logic sounds promising! I see it being really useful for large Hadoop applications and ML models. Would love to hear how others are adapting their Kubernetes setups for such workloads. Especially, I'm curious about experiences with large images.

Image Volumes are a game changer for me! They let you separate your core code from additional files, which is great for LLMs, plugins, or static assets. Just keep in mind that this feature requires the upcoming containerd 2.1 release. Currently, it's still new with cri-o, and many managed clusters are on containerd 1.7.x, so it might take time before it's widely usable.