There are multiple scenarios where migration is the best or most feasible approach. For example, you:
- Have capacity on a new cluster.
- Are doing a hardware refresh such as for greater capacity or performance.
- Do not want to disturb existing workloads.
- Can move workloads one at a time.
- Do not want any downtime.
- Have an existing instance of CDH or HDP that does not have a direct upgrade supported.
There are several tools available, including:
- Workload XM
- Enables you to migrate or shift workloads after analyzing them, and helps you move workloads one by one.
- Replication Manager
- Helps you replicate and copy the data and metadata.
The migration process is to:
- Configure the new cluster.
- Identify the candidate workloads.
- Copy the data, metadata, and policy.
- Migrate and test the workloads.
- Promote the new cluster and workloads into production.
- Decommission the legacy cluster.
- Depending on hardware compatibility, add the nodes to the new CDP Private Cloud Base cluster for additional capacity.
This process enables you to perform a rolling migration of the cluster, by both:
- Deploying new hardware.
- Repurposing existing hardware by gradually migrating data and workloads to the new cluster.
An overview of the migration process is shown in Migration to CDP Private Cloud Base.