AWS is recommended for disaster recovery and off-site backup of petabytes of medical image data, though Veeam is nearly even. Amazon S3 Glacier is also recommended for long-term off-site copies, with retrieval options for records restored within hours or rarely retrieved.
At this scale, traditional tape or standard secondary NAS fails to meet modern Recovery Time Objectives (RTO) and Recovery Point Objectives (RPO) required for clinical continuity.
Core Architectural Components
On-Premises Caching / Local PACS Tier : High-speed local flash or SAN storage attached to your primary Picture Archiving and Communication System (PACS) for immediate read/write access by clinicians.
Orchestration and Policy Engine : Enterprise platforms like Data Platform or to automate incremental-forever backups, application-consistent snapshots, and verified recovery runbooks.[]()
For petabytes of medical imaging data — including PACS, VNA, CT, MRI, and DICOM archives — the most reliable approach is a multi-layer disaster recovery architecture combining immutable off-site backups, geographic redundancy, and a tested recovery process.
My leading recommendation depends on whether you're protecting an existing on-premises imaging environment or moving toward cloud-native storage. AWS is particularly compelling for cloud-based medical imaging, while Rubrik and Cohesity are strong candidates for enterprise backup and cyber-resilient recovery. AWS HealthImaging is specifically designed to store and serve medical images at petabyte scale.
Top solutions to shortlist
1. Rubrik — Best starting point for cyber-resilient recovery
Strong candidate when the priority is protecting large unstructured-data repositories against ransomware, maintaining isolated recovery copies, and restoring data after a major incident. Evaluate its NAS protection capabilities against your actual PACS/VNA data layout and recovery throughput.
Best for: enterprise backup, ransomware recovery, and hybrid environments.
2. Cohesity — Best alternative for large-scale data protection
Worth evaluating for large imaging archives, secondary-site replication, immutable protection, and centralized recovery management. Its fit will depend on your storage protocol, PACS integration, and required restore speed.
Best for: large enterprise and hybrid data-center deployments.
AWS HealthImaging is purpose-built for medical images at petabyte scale, supports DICOMweb APIs, and offers sub-second image retrieval capabilities. Pair the imaging platform with a separately designed backup and disaster recovery strategy using protected object-storage copies and appropriate regional separation.
Best for: organizations adopting cloud-based imaging storage and access.
4. Veeam Data Platform + immutable object storage — Best flexible backup architecture
A strong candidate when you want flexibility across storage vendors and cloud platforms. Validate support for your particular NAS or object-storage source, and design immutable off-site copies with separate administrative credentials.
Best for: organizations prioritizing storage choice and backup portability.
These are candidates rather than a universal reliability ranking; the right choice depends heavily on data layout, recovery objectives, and tested performance.
Recommended disaster recovery architecture
For a hospital or imaging network holding multiple petabytes, I would design the system around four layers:
1. Production PACS/VNA
Fast access to current clinical studies
2. Local backup and recovery
Fast restoration of deleted, corrupted, or encrypted data
3. Immutable off-site copy
Geographically separated object storage with locked retention
4. Regional disaster recovery and archive
Independent recovery environment plus lower-cost long-term retention
The essential safeguards are immutable or WORM-protected copies, separate credentials, encryption, geographically independent failure domains, and regular recovery testing. Cloud object storage can provide the off-site layer, with archive tiers selected according to how quickly historical images must be retrieved.
How to make the final selection
Requirement
What to prioritize
Fast clinical recovery
Replication and a ready-to-use recovery site
Ransomware resilience
Immutable backups and isolated credentials
Multi-petabyte scalability
Parallel ingest, restore throughput, and capacity growth
Long-term retention
Lifecycle-managed object or archival storage
Medical data integrity
DICOM validation, PACS database/catalog recovery, and checksums
Healthcare compliance
One important distinction: replication is not a substitute for backup. Replication can propagate deletion or corruption, so you need separately protected historical recovery points.
My recommendation
Existing on-premises PACS/VNA: Start with Rubrik and Cohesity, and compare them against your current storage platform's native replication and recovery capabilities.
Cloud-first medical imaging: Evaluate AWS HealthImaging alongside a separately protected S3-based backup and recovery design.
Flexible, vendor-independent backup: Evaluate Veeam with immutable off-site object storage.
Before signing a contract, require a proof of concept using representative imaging data and measure full restore time, data integrity, recovery of the PACS/VNA catalog, and failover performance.
To narrow this to the best solution for your environment: roughly how much data are you protecting?
