Update Description
The I/O passthrough tool optimizes the object storage device (OSD) storage I/O process in balanced configuration. It applies only to the OSDs that use the BlueStore storage engine and whose data and DB partitions are located on different drives.
OSDs may be added or modified in Ceph clusters. The tool needs to obtain information of the OSDs in the clusters. Therefore, the OSDs must be deployed in the default official directory /var/lib/ceph/osd. If the OSDs are deployed in other directories, the tool will not work.
The storage configurations can be classified into the following types:
- Balanced: The cluster uses solid-state drives (SSDs) and hard disk drives (HDDs) to deploy OSDs. SSDs are used to store OSD metadata (DB) and log data (WAL), and HDDs are used to store real data (data).
- All-flash: The cluster uses SSDs to deploy OSDs. The data, DB, and WAL of the OSDs are stored in the SSDs.
- Cold storage: The cluster uses HDDs to deploy OSDs. The data, DB, and WAL of the OSDs are stored in the HDDs.
- Bcache: The SSDs and HDDs are bound as a Bcache partition. The SSDs are used as the cache of the HDDs. The cluster uses the Bcache partition to deploy OSDs. The data, DB, and WAL of the OSDs are stored in the Bcache partition.
Currently, the I/O passthrough tool applies only to the balanced configuration.
Parent topic: Release Notes