I got an email from Google Cloud Platform today entitled:
[Action Required] Upgrade to Artifact Registry before March 18, 2025
This is not the first time Google has discontinued a product I use. They gave full year of lead time but I knew I would forget about it before then. I decided to look into why they are making the change and what would I have to do to make everything work.
Why?
I first checked out a blog on this topic from Google. The main driver behind the name change it seems is that Artifact Registry can manage more than just containers, such as OS and language packages. This wasn’t a feature I was missing, but I’m sure other people were and I may later. It also goes on to explain that now you can create regional, rather than multi-regional, repositories. Both Container Registry and Artifact Registry are priced based on storage costs and network egress. Apparently, per unit storage costs are going to be higher in Artifact Registry than the were in Container Registry. However, if your Container Registry to Artifact Registry migration will result in having artifacts stored and used in the same region, there will be no egress charges. I think this means that most of my use cases will just cost more.
The good news is Artifact Registry integrates with Cloud IAM for more fine grained security. The ability to restrict access to regions, repositories and environments as opposed to just a project level is worth it for me. The other two changes that I mentioned are improvements, but they wouldn’t justify too much migration effort for the projects I currently have. However, fine-grained security in enterprise environments is always welcome. Assuming there isn’t too much effort involved.
How?
Google provides documentation and tooling to help the Container Registry to Artifact Registry migration. Usually I can provide some insights into practical workarounds for common issues, but I didn’t have any problems. I migrated from the standard repositories because I’m not using the gcr.io domain. Maybe there might have been more drama if I was using customer-managed keys or some other configuration that I would have to address. I probably could have even used their automated process. I was able to just copy images from the Container Registry to the Artifact Registry manually.
When?
You can see the timelines. This is a phased approach. Keep in mind, a phase may seem short or long depending on the specifics of your Container Registry to Artifact Registry migration process and your current backlog. You can plan out you migration by following a guide GCP provides. If you don’t already have a good idea of your current Container Registry ecosystem, there’s no time like the present to get up to speed. If you are going to use their automated process, there are some permissions that need to be requested and some other implementation details that will require an administrator.
What’s Next?
Many organization have a lot invested in GCP’s Container Registry and will need to take a controlled, measured approach to this migration. This is not optional. And it may not be simple. Perficient’s Cloud Consulting Services can help you evaluate and implement your Container Registry to Artifact Registry migration, regardless of the complexity.
