Unifying Database Backup and Disaster Recovery with Copy Data Virtualization

Background

Virtualization of servers, storage and networks has matured, but copying production data has not. Enterprises often keep 10‑20 copies of the same database for backup, disaster recovery, testing, compliance and analytics, which creates high production load, storage redundancy and management complexity.

Copy Data Virtualization (CDV)

CDV solves three core problems:

Provides a single acquisition interface that minimizes impact on production systems.

Consolidates storage into two pools – a snapshot pool for recent full snapshots and a deduplication pool for compressed, deduplicated long‑term data – reducing redundancy and cost.

Enables on‑demand use of copies for backup, disaster recovery (DR), development, testing and analytics.

Architecture Overview

Data acquisition : Supports IN‑BAND and OUT‑OF‑BAND methods; OUT‑OF‑BAND is recommended to keep production impact minimal. Oracle data (datafiles, controlfiles, spfiles, archivelogs) are captured via RMAN; VMware virtual disks are obtained through the vStorage API.

Data storage : Two logical pools. The snapshot pool holds recent full snapshots generated after each incremental backup. The deduplication pool stores compressed, deduplicated snapshots for long‑term retention.

Remote replication : An optional DR site can host a synchronized CDV instance. Replication may be synchronous or asynchronous over IP or Fibre Channel.

Data usage : Any snapshot in the snapshot pool can be mounted instantly on the original host or a remote host, providing minute‑level recovery and rapid provisioning of development/test environments.

Oracle Backup and Recovery Workflow

Mount a virtual volume from the CDV platform as a staging disk on the database server.

Use Oracle RMAN IMAGE COPY to copy the full data set onto the staging disk and perform incremental merge operations.

Unmount the staging disk from the database service.

Create a snapshot of the staging disk in the CDV snapshot pool; the snapshot is application‑consistent.

LogSmart automates archival log backups according to defined intervals, allowing point‑in‑time mounts with minute‑level granularity.

Key Technical Benefits

Acquisition efficiency : One full copy followed by continuous incremental updates dramatically reduces production I/O.

Copy virtualization : A single physical copy can spawn unlimited virtual copies, each readable and writable for independent purposes.

Mount‑based access : Snapshots can be mounted directly, enabling instant read/write access, rapid DR, and fast creation of test environments.

Typical Use Cases

Backup of large databases (e.g., single instances >20 TB) with always‑incremental backups and instant mount recovery.

Backup of VMware VM pools, providing storage‑level fault tolerance and immediate VM restoration.

Emergency database recovery with minute‑level RTO.

Rapid provisioning of development and testing environments that mirror production data.

Integrated backup‑DR solution that serves both protection and testing from a single platform.

Technical Q&A Highlights

Data‑loss window depends on the backup SLA and frequency of archival log backups; with timely log backups, loss can be limited to minutes.

Remote DR site I/O performance may be lower, especially over long distances; choice of synchronous vs asynchronous replication affects latency.

Compared with Data Guard/OGG, CDV relies on archived logs to avoid data loss; both approaches can achieve zero‑loss recovery if logs are captured promptly.

Oracle acquisition requires an Actifio agent on the source host; VMware acquisition does not require an agent because it uses the vStorage API.

Replication links can be IP networks or Fibre Channel; synchronous replication is possible when using IN‑BAND acquisition.

Mounted snapshots are fully read/write and can be mounted on multiple hosts simultaneously.

Merge duration is proportional to data size; RMAN logs contain detailed merge timing.

The Actifio agent captures data at the file‑system layer.