With Dgraph Cloud going EOL on 9/30/25, I need to get self-hosted up and running asap or risk major service disruption.
I likely will use AWS EKS and the dgraph helm chart
I believe this should work but I have questions about using persistent volumes as the underlying data storage.
- Persistent volumes are usually tied to the cluster lifecycle, so if I have an eks cluster running my dgraph deployment, then I need to create a new eks cluster for it (e.g. using a blue/green deployment), how can the underlying volumes be maintained and for some amount of time be accessible to both the old and new clusters running my dgraph workloads?
- For multi-node dgraph deployments recommended for production, if there are dgraph zeros and alphas running on different eks nodes, the underlying file storage can’t use EBS since pods on different nodes will need access. Is this correct? If so, what is recommended storage type for this scenario?