Chapter 3. Hardware Considerations
The ideal target environment for installations using the Console is production Ceph deployments, specifically hardware environments that conform to the best practices established by Ceph storage experts at Red Hat. For example, it is recommended that the Ceph OSD nodes have a small number of solid state disks (SSDs) to serve as shared journal devices.
The Ansible-based installer embedded with the Storage Console auto-detects these devices and make intelligent layout decisions for journal placement and placement group (PG) counts given the hardware at hand. If no SSD devices are present, the installer uses half of the OSD devices for journals and half for OSDs. This might not be optimal for some use cases where storage density is the primary concern, over performance.
Red Hat Storage Console supports importing clusters configured with journals and OSDs on dedicated physical disks. Currently, the Storage console does not support collocated journals and OSDs on the same physical disk. If a user imports a cluster with journals and OSDs collocated on the same physical disk, there can be the following undesirable consequences:
- Incorrect journal size and journal path shown in the OSD tab of the Host details view
- Provisioning operations and automatic cluster expansion operations by the Storage Console would be performed using journals and OSDs on dedicated physical disks and would not maintain the existing colocation configuration
Therefore, owing to these consequences, importing clusters with colocated journals and OSDs is not recommended.