We've used Veritas NetBackup on-premises for several years, but the licensing and infrastructure costs are becoming difficult to justify. We're evaluating alternatives, mainly for VMware workloads, and want to understand the real cost and scalability differences.
For an initial model, we're considering 10 VMs at roughly 200 GB each. Five would need one year of retention, while the other five would be retained for 30 days. That gives us about 2 TB of initial backup data with a relatively low daily change rate of approximately 0.1%.
We're comparing an AWS-native design using AWS Backup, EBS snapshots, and S3 lifecycle tiers against Veeam or similar third-party products, as well as traditional enterprise platforms such as NetBackup. We're particularly interested in how long-term retention, storage growth, licensing, infrastructure, data transfer or egress, restore performance, disaster-recovery testing, and operational effort affect the total cost.
AWS appears simpler on paper, but we're unsure whether that remains true as retention and restore requirements increase. Veeam offers more flexibility, but we don't know whether it actually lowers costs or simply moves them into licensing, storage, and management overhead.
What cost drivers have mattered most in real deployments, and have third-party tools justified their additional complexity compared with an AWS-native or legacy approach?
4 Answers
I wouldn’t compare these options using storage cost alone. Licensing is only part of the picture; restore speed, recovery testing, administration, support, and the cost of a failed recovery can matter more than the monthly repository bill. Make sure the model includes the infrastructure and staff time needed to operate each design.
The biggest AWS-specific concern is recovery and data movement. Check whether the service deduplicates or compresses data in the way you expect, then model restore traffic, cross-region transfers, and any NAT gateway charges. A regular disaster-recovery test can generate much more egress than normal backups. Also compare how easily each option supports application-aware backups, granular restores, backup verification, replication, and booting a VM directly from backup.
For an on-premises VMware environment, a common practical design is Veeam with a local repository for fast restores and immutable copies in object storage such as Wasabi or Backblaze B2. That gives you quicker recovery for ordinary incidents while keeping an off-site, tamper-resistant copy for ransomware or site loss. It may be less expensive than putting everything directly into AWS, depending on storage pricing and your restore pattern.
That’s the kind of split design we’re considering. The local repository and object-storage copy would need to be included in the comparison rather than treating the Veeam license as the only extra cost.
Think beyond where the backup files live. If the primary site is unavailable for weeks, simply having data in object storage may not be enough—you also need somewhere to run the workloads. A managed backup or DR provider can cost more per month but may include recovery infrastructure, verification, bandwidth, and access to support engineers. Those operational and business-continuity costs are difficult to quantify, but they often determine whether the solution is actually useful during a major incident.
Exactly. A low backup-storage estimate can be misleading if a full-site recovery requires pulling terabytes over a WAN. The recovery runbook and test frequency should be part of the financial model.