How Dell EMC RecoverPoint Achieves Millisecond‑Level Data Recovery for Enterprise Continuity

Test Environment

The lab built by the Enterprise Research Institute used the following components:

PowerEdge R840 4‑socket servers running Oracle RAC.

Dell EMC SC7020 primary storage (Site A) and secondary SCv3020 storage (Site B).

Dell EMC VPLEX for virtualized block‑level data services across sites.

Dell EMC RecoverPoint for continuous data protection and point‑in‑time recovery, with backup copies stored on a VNX7600 appliance at Site C.

Swingbench as a synthetic workload generator to keep the Oracle database under constant load during the tests.

The topology formed a “two‑site‑three‑center” architecture: Site A (production), Site B (same‑city disaster‑recovery), and Site C (remote backup).

Test Scenarios

Test 1 – Production storage failure and automatic failover

Simulated a hard‑failure of the SC7020 array at Site A.

VPLEX detected the loss of the storage and automatically redirected all I/O to the replica at Site B.

Oracle RAC continued to serve requests; transaction‑per‑second (TPS) and transaction‑per‑minute (TPM) remained stable.

The only observable impact was a minor “saw‑tooth” pattern in CPU usage on the R840 host.

Test 2 – Database corruption, loss, or accidental deletion

After the storage failure, the SC7020 array was repaired and brought back online while Swingbench continued to generate load.

VPLEX automatically re‑established the primary path to Site A and synchronized data written during the outage.

RecoverPoint performed a full restore of the Oracle database from the remote copy at Site C within a few minutes, regardless of the data volume.

Test 3 – Data tampering and point‑in‑time recovery

Engineers manually deleted the system01.dbf file on the Oracle host, causing the instance to fail to start.

RecoverPoint was instructed to roll the database forward to a specific timestamp captured on the remote replica.

The point‑in‑time restore completed in a few minutes and the database returned to the pre‑tampering state.

Test 4 – Hacker‑induced data alteration and targeted recovery

A simulated intrusion granted admin rights and modified several database rows.

RecoverPoint was used to restore the database to the exact point just before the malicious changes.

The operation demonstrated millisecond‑level granularity for selective recovery.

Backup Granularity Options

Dynamic : Backups are created based on actual write activity, adjusting frequency automatically.

Fixed per Second : A snapshot is taken every second, independent of write volume.

Fixed per Write : A snapshot is taken for each write operation, providing the finest granularity at higher storage cost.

Replication Modes

Synchronous replication for low‑latency, same‑city “dual‑active” configurations.

Asynchronous replication for long‑distance, remote‑backup scenarios where higher latency is acceptable.

Key Findings

Automatic failover and failback are fully transparent to the Oracle workload; no manual intervention is required.

RecoverPoint delivers millisecond‑level point‑in‑time recovery, enabling precise restoration after corruption or malicious alteration.

The three backup granularity modes allow enterprises to balance storage cost against acceptable data‑loss windows (RPO).

Both synchronous and asynchronous replication support flexible “two‑site‑three‑center” designs, reducing RTO and RPO across diverse failure domains.