Currently the StorageID is used to identify both the Datanode and the storage. This works because we treat all storages attached to the Datanode as a single storage unit.
We plan to replace the StorageID with two independent IDs:
- Datanode UUID - this will be a string that uniquely identifies each Datanode in the cluster. For upgraded clusters, the Datanode UUID will be the StorageID prior to the upgrade. i.e. we will reuse the prior Storage ID.
- Storage UUID - Each storage attached to the datanode will be identified by a UUID.