2.1.5 Improved Data Durability with Exascale Delta Tracking

Oracle Exadata System Software release 25.2.0 introduces optimizations for handling multiple failures impacting all mirrored data copies on Exascale.

Previously, Exascale responded to disk failure by forcibly dropping the associated pool disk. This approach enabled the storage pool to quickly start a rebalance operation and restore full redundancy, but it provided no opportunity to rescue the failed disk. Also, in the rare case where multiple failures impact all mirrored data copies, forcibly dropping all of the corresponding pool disks causes data loss within the storage pool.

Starting with Oracle Exadata System Software release 25.2.0, Exascale immediately responds to a disk failure by taking the affected pool disks offline rather than dropping them. If any mirror copies remain intact, Exascale simultaneously tracks new writes as deltas on the remaining online disks and begins rebuilding full data redundancy on other available disks.

If a failed disk comes back online before full redundancy is restored, it is quickly synchronized using the stored deltas, minimizing impact and enhancing data durability.

Data durability is further improved in the rare event that all mirrors are affected by multiple simultaneous failures. In such cases, the affected storage pool ring goes offline but can be restored by bringing any one of the failed disks back online.