Determine how to store metadata
Application developers should use an external database to store object metadata. Client-side metadata operations can involve complex and fast queries that are demanding on subsystems common to many database technologies.
Most databases index fields that are commonly searched. IBM Cloud Object Storage System™ storage is paired successfully with numerous database technologies, including many open- and closed-source RDBMS, key-value stores, and NoSQL cloud document systems.
All objects that are written to the system are immutable by design.
To replace or update the content of an object, a new object must be written again and the old object deleted. Typically, client applications store the OID returned from the system as a column in a database table.
When object content is updated, this OID is updated in that database table. Applications store object metadata as other columns within the application database. It allows the application to indexing and query the metadata.
- Database replication
- Replicate the database if the size of metadata is insignificant compared to content data size.
- Metadata dispersal
- Store bundled and non-searchable metadata in an SOV either:
- Covariant metadata dispersal
- Updates metadata only when the object updates.
- Independent metadata dispersal
- Updates metadata whenever an update is needed.