IBM Cloud Object Storage provides the storage for workspaces. When you create a project, catalog, or deployment space workspace, a dedicated storage bucket is created to store the assets and files for that workspace.
If you have more than one IBM Cloud Object Storage instance, you must select an instance when you create a workspace. You can also connect to IBM Cloud Object Storage as a data source. See IBM Cloud Object Storage connection.
What is stored in Cloud Object Storage buckets for workspaces
Items that are stored in a Cloud Object Storage bucket for a workspace are a combination of user files that you explicitly add, files for assets that you create, and system files that are created automatically.
The following types of items are stored in the storage buckets that are associated with workspaces:
- Files that you upload into the workspace, such as data files
- Files for assets that you create, such as notebooks, models, or flows
- Files that you create as part of notebook processing, such as data tables
- System files that are created automatically, such as log files
For example, when you run an AutoAI experiment, the following types of files are created:
- Binary files for each model-candidate pipeline
- Auto-generated notebooks (one per pipeline and one per model)
- CSV files that contain plot data for metric visualizations
- Metadata files used to display the experiment in the user interface
When you delete an asset in a workspace, all the files that are associated with that asset are deleted from the storage bucket.
Security and resiliency of Cloud Object Storage
Each workspace has its own bucket. From within a workspace, only the workspace collaborators can access items in the bucket. From IBM Cloud, only users who have the IAM role of Manager or Writer for the IBM Cloud Object Storage service can manage the bucket. See Provide access to Cloud Object Storage.
All data that is stored in IBM Cloud Object Storage is encrypted and erasure-coded. You can optionally encrypt the Cloud Object Storage instance that you use for workspaces with your own key. See Encrypt IBM Cloud Object Storage with your own key. The Locations in each user's Profile must include the Global location to allow access to Cloud Object Storage.
Cloud Object Storage is resilient by dispersing data across three locations. When you create a workspace, the Cloud Object Storage bucket defaults to Regional resiliency. Regional buckets distribute data across several data centers that are within the same metropolitan area. If one of these data centers suffers an outage or destruction, availability and performance are not affected.
Learn more
- Setting up IBM Cloud Object Storage
- IBM Cloud docs: Getting started with IBM Cloud Object Storage
- IBM Cloud docs: Endpoints and storage locations
- Managing storage space in Cloud Object Storage
- Troubleshooting Cloud Object Storage for projects
Parent topic: Overview of Cloud Pak for Data as a Service