What is sparse disk group in Exadata?
In the Exadata ecosystem, a Sparse Disk Group is the architectural magic behind ultra-fast, space-efficient Snapshots.
If you’ve ever needed to create a dozen copies of a 10TB production database for testing—and needed them ready in minutes without buying 120TB of extra storage—the Sparse Disk Group is your solution.
A traditional "Full" Disk Group stores every byte of data. A Sparse Disk Group, however, only stores the changes (deltas) made to a base image.
It acts as a "Read-Only Redirector." When you create a snapshot, Exadata doesn't copy the data; it creates a pointer to a Test Master (a read-only version of your source database).
The Sparse Disk Group utilizes a Redirect-on-Write mechanism. Here is the lifecycle of a data block in this setup:
The Read: When a developer queries a snapshot, the system checks the Sparse Disk Group. If the block hasn't been changed, it transparently reads the original block from the Test Master.
The Write: When the developer updates a row, the new version of that block is written into the Sparse Disk Group.
The Result: The original Test Master remains untouched (read-only), while the Sparse Disk Group grows only by the size of the new modifications.
The primary driver for Sparse Disk Groups is DevOps agility and Storage Economics.
Near-Instant Cloning: You can "clone" a multi-terabyte database in seconds because no physical data is being moved.
Massive Space Savings: You can host 20+ snapshots of the same production database using only a fraction of the storage, as each developer only stores their unique changes.
Hardware Acceleration: Unlike generic storage snapshots, Exadata Sparse Snapshots still benefit from Smart Scans and Columnar Caching.
| Feature | Full Disk Group | Sparse Disk Group |
| Data Content | Complete copy of all blocks. | Only changed blocks (deltas). |
| Creation Time | Hours/Days (Physical copy). | Seconds (Metadata only). |
| Primary Use | Production / DR. | Development / Testing / QA. |
| Storage Cost | 1:1 ratio. | Often < 10% of the original size. |
To use Sparse Disk Groups, you first create a Test Master. This is a full, read-only copy of your database (usually created from a backup or a standby).
Once the Test Master is flagged as read-only, you can "spawn" as many sparse snapshots as your CPU can handle. If a developer accidentally deletes a table in their snapshot, only their sparse storage is affected—the Test Master and all other snapshots remain perfectly intact.
Sparse Disk Groups transform Exadata from a high-performance database machine into a private cloud platform. They allow teams to iterate faster by giving every developer their own "production-sized" playground without the production-sized storage bill.