Repository Types
Recommendations for general repositories are outlined in VBR best practices.
Yet, the need for scalability and multitenancy in IaaS data protection brings about specific considerations that affect the choice of repository type.
As immutable storage is key in data protection scenario, it’s advised to use on-premises S3-compatible object storage or VHR as a landing zone, expanding it with capacity and archive tiers.
It’s advisable to follow best practices by utilizing SOBR offloading via the capacity tier (rather then using Backup Copy to S3). The fundamental reasons are:
- It removes the necessity for configuring manual Backup Copy jobs.
- It offers clear visibility of tenants’ backups across all tiers during self-service restore.
- Optimizing restore performance and cost through the reuse of matching blocks from on-premises repositories.
Since VBR servers cannot share components, opting for VHR would exponentially increase the number of required physical repositories. This is illustrated in the example below.
That’s why in environments with multiple active sites and a need for cross-site copies, it’s highly recommended to use S3-compatible object storage. The adoption of S3-compatible object storage significantly streamlines the architecture, offering a more efficient solution.
Scale-Out Backup Repository
Due to complications in obtaining accurate repository consumption per tenant, one widely used approach is to use a dedicated repository for each customer.
Keep in mind that, according to internal tests and field experience, the maximum number of SOBR (Scale-Out Backup Repository) repositories per VBR installation is 100. Significant performance drops can be expected after exceeding this threshold.
This limit applies to the number of SOBRs and does not affect the number of extents. The type of repository and the use of a capacity tier do not impact this number as well.