Define content data and metadata
Conceptually, a client application can view an object as both its content data and any associated metadata.
Objects that are written to an SOV cannot be read unless a threshold number of slices are
recovered. It means that objects cannot be searched. The SOV type is suitable for storing the
content data portion of an object.
| Metadata | Content data | |
|---|---|---|
| Scaling | Small relative size, scale up scale or partitioning | Massive relative size, out slice servers |
| Search | Indexed, fast search | Binary blob, no search |
| Reliability | Replication | Replication |
Typically, metadata is smaller in size and must be indexed and searchable. A database is better suited to store metadata because it can be indexed and searched. A database can make data more reliable by replicating it. Any hierarchy, file names, or other such metadata should be stored with the object metadata in an application database.
Content data is often of limited use without its associated metadata. A developer needs to ensure
that metadata is stored as reliably as the content data stored on a SOV. Two approaches to
accomplish it are available.
- Use built-in replication features of the application database.
- Back up a copy of the metadata to the IBM Cloud Object Storage System™.
A metadata backup cannot be searched so an indexed copy of the metadata must be in the application database.