“What’s the best software to back up Amazon S3?” My short answer is that it depends on where the copy has to live, because that one decision sorts Amazon S3 backup software faster than any feature grid.
The quick version:
For cloud-first AWS, Azure, and Google Cloud environments, Eon is the best Amazon S3 backup software because it discovers buckets by itself and restores anything from one object to a full bucket into any Region, without leaning on S3 Versioning.
Most AWS teams keep AWS Backup as their native foundation, and Druva fits teams who want a managed SaaS to hold copies outside their AWS organization. Clumio by Commvault suits AWS-only shops chasing 15-minute recovery points.
Veeam and Rubrik make sense when a data-center deployment already anchors the estate, and MSP360 is the budget way to copy a few buckets into another cloud.
| Software | Where the backup copy lives | Point-in-time restore | Pricing model |
| Eon | Immutable vault apart from the source account | Any point, no versioning needed | Usage-based, per GB/month |
| AWS Backup | Your vault, or a logically air-gapped vault | 35 days (continuous) | Per GB-month plus requests |
| Druva | Druva Cloud, outside your AWS organization | Latest or a chosen recovery point | Per TB/month after dedup |
| Clumio by Commvault | SecureVault outside the account, or in-account Backtrack | With continuous backup | Per GiB plus per-object fee |
| Veeam | A repository you run | Rollback to a restore point | Per workload, 500 GB per instance |
| Rubrik | S3 in the account and Region you pick | Snapshot-based | Quote only |
| MSP360 | Another cloud account | Per-file versions | Per computer, per year |
Where should an S3 backup copy live?
Every tool here answers the location question differently, and the answer decides what survives a stolen admin key.
- In the source account: The copy sits next to the data it protects, which keeps setup simple. AWS Backup’s standard vaults and Clumio’s Backtrack work this way, so anyone holding production credentials stands near the backups too.
- In a separate vault account: A logically air-gapped vault holds the copy apart from production, so a compromised source account can’t reach it. AWS Backup’s logically air-gapped vaults work this way, and so does the #1 pick below.
- In a vendor’s cloud: Druva keeps copies in its own cloud outside your AWS organization, and Rubrik offers a managed model that moves data outside the organization by design. You get isolation without running anything, in someone else’s environment.
My bias for production buckets is the middle option, because a copy that production credentials can’t open is the one still standing after a bad day.
1. Eon: cloud-native S3 protection for cloud-first AWS, Azure, and Google Cloud environments
Built by the team behind CloudEndure, this cloud data protection and recovery platform starts its S3 coverage with a problem most teams underestimate: knowing which buckets exist.
Pick it if: your buckets are spread over dozens of AWS accounts, Azure Blob Storage and Google Cloud Storage may be in the mix, and some of those buckets have never had versioning enabled (you know the ones).
Under the hood: Onboarding takes a single read-only IAM role, and there are no agents, appliances, or clusters to deploy. Cloud Backup Posture Management (CBPM), the category the company started, then keeps finding and classifying buckets as accounts change and applies policy with no tagging.
Each object is captured individually, and after the first pass only new and changed objects are stored, so versioning never enters into it. Recovery targets an entire bucket, one prefix, or a single object at any past point, straight from backup storage into whichever Region you need.
Global Search covers every snapshot without an index to set up, so the overwritten object a developer is panicking about shows up in seconds. The vault is immutable, logically air-gapped, and separated from the source account, and access to it is time-bound, which keeps production keys out.
Ransomware detection (add-on) checks every incoming backup for version flooding, mass overwrites, and staged encryption, and it marks the most recent clean recovery point. For long retention, the S3 Vault Cold Tier keeps older copies at lower cost.
What you’ll pay (as of October 2026): Pricing is usage-based, billed per GB per month on protected storage, with no tier ladder and flexible spending commitments. Vault-side global deduplication and incremental capture drive the company’s up to 50% lower cost figure for S3 backup storage.
Retrieval, restore, and egress carry no fees, while ransomware protection sits in a separately priced package.
Limits: Its on-prem support is limited, so an estate that’s still mostly on-prem needs a separate tool for the data center. Expect a quote (there’s no public list price), and if you only run a few buckets in a single account, AWS Backup on its own will probably cover you.
2. AWS Backup: the native baseline for versioned buckets
I’d turn AWS Backup on before shopping for anything else, since it’s already in your account and uses the same plans as your EC2, RDS, and DynamoDB resources.
Pick it if: centralized native policy is the goal, set once through AWS Organizations and checked by Backup Audit Manager against your own controls, and versioning is already on for the buckets you care about.
Under the hood: Point-in-time recovery from continuous backups covers the last 35 days, reaching as close as 15 minutes before now. Scheduled snapshots can fire as often as hourly or as rarely as monthly, with retention stretching to 99 years.
S3 Versioning is a prerequisite on every protected bucket, and continuous mode leans on EventBridge. Recovery points can be copied across accounts and Regions, and a logically air-gapped vault keeps them in an account AWS Backup owns, under Vault Lock compliance mode.
This year brought two useful upgrades: GuardDuty malware scans for S3 continuous backups in May (a separate GuardDuty charge), and direct read access to backup data over standard S3 APIs in August.
What you’ll pay (as of October 2026): List prices in us-east-1 are $0.05 per GB-month for warm storage, $0.035 once data moves to the low-cost warm tier after 60 days, and $0.0575 inside a logically air-gapped vault. Restores run $0.02 per GB.
EventBridge events and S3 requests add to the bill, and any object smaller than 128 KB is billed as if it were 128 KB (bad news for buckets full of tiny files).
Limits: A single item-level restore job handles at most five objects or folders, and copied continuous backups give up their point-in-time history. AWS’s docs also say that switching off a bucket’s EventBridge notifications breaks continuous backup without an alert or failed job to show for it.
3. Druva: managed SaaS that keeps S3 copies outside your AWS organization
Druva covers S3 from the same fully managed SaaS that protects endpoints and SaaS apps, a good match for teams who’d happily never patch a backup server again.
Pick it if: you’d rather a vendor hold the S3 copy, you want nothing to install, and your objects live in S3 Standard or S3 Intelligent-Tiering.
Under the hood: Buckets go into backup sets, and each set gets a policy for schedule and retention. Druva keeps the resulting snapshots in Druva Cloud, outside your AWS organization, so the copy sits in a different security domain from day one.
Recovery can be filtered by object size, type, or time, and it returns objects, whole buckets, or metadata. Druva’s August 2025 S3 launch added restores into another AWS account or Region, handy when the account that failed is the one you’d normally restore to.
What you’ll pay (as of October 2026): Business, Enterprise, and Elite plans are priced per TB per month after deduplication, and Druva’s pricing page shows the tiers without dollar amounts. Capacity is bought as prepaid credits, each worth one TB-month.
Per Druva’s credits help article (September 2026), backup operations pause when credit usage hits 110%, so put someone in charge of watching it.
Limits: Supported storage classes in Druva’s S3 docs are S3 Standard and Intelligent-Tiering, and SSE-C encrypted objects need a different route. As of October 2026, its plan page places cross-Region backup and Data Lock on Enterprise and Elite.
The architectural trade is the flip side of the isolation: with the copy in a vendor-run cloud, every recovery goes through Druva’s console and workflow.
4. Clumio by Commvault: tight recovery points for very large AWS-only buckets
Clumio, which Commvault acquired in 2024, is a managed service with two flavors of S3 protection, and they sit on opposite sides of the isolation question.
Pick it if: your estate is AWS-only, individual buckets are huge, and recovery-point tightness beats bill simplicity on your priority list.
Under the hood: SecureVault keeps an incremental second copy beyond the production account. Backtrack does in-account rollback and needs S3 Versioning, while continuous mode polls every 15 minutes and syncs after more than 32 changes or hourly at the latest.
Restores go to the source bucket, a fresh bucket, or one in a different account and Region, at object, prefix, or bucket scope, with the current version or all of them.
SecureVault Standard also exposes every point-in-time backup over S3 APIs, so you can browse before committing to a restore. Commvault says it has tested single buckets holding 80 billion objects and 40 PB.
What you’ll pay (as of October 2026): Storage in SecureVault Standard costs $0.025 per GiB-month, plus a $1.50 monthly fee per million managed objects, and SecureVault Archive cuts storage to $0.010 per GiB-month with that same object fee.
Restores bring their own meter: $0.04 per GiB for direct access, $0.09 per GiB retrieved, $0.025 for every 1,000 API requests, plus transfer out (so model a bad month before you have one).
Limits: Glacier and Glacier Deep Archive appear as unsupported in Clumio’s documentation, and objects in Intelligent-Tiering’s archive tiers fail backup. Backtrack lives inside the account it protects, and Archive restores take 24 to 48 hours to start.
5. Veeam: S3 buckets inside an existing Veeam Backup & Replication server
Veeam’s home turf is the VM and hypervisor estate, and its S3 support reads best as an extension of that footprint.
Pick it if: your data center already runs on Veeam, and you’d like buckets to share its backup server, policies, and repositories.
Under the hood: Bucket protection runs as a Veeam Backup & Replication v13 job type for object storage, where a proxy pulls objects and writes them to whichever repository you point it at, on-prem included.
You can bring back an entire bucket, every object in one restore point, or several versions of an object, either in place or somewhere new. Point-in-time rollback touches only the objects that changed (a nice way to keep a large recovery small), and an immutable repository supplies the immutability.
What you’ll pay (as of October 2026): Veeam Data Platform meters licenses by workload, and each license instance covers 500 GB of protected object data. Expect a quote for dollar figures.
Then add the backup server, proxies, and repository storage you run.
Limits: SSE-C objects, Glacier Flexible Retrieval, and Glacier Deep Archive fall outside object storage backup per Veeam’s docs, which also note that prefix restores can’t come from a long-term repository.
Every component is yours to run, which a data-center team already knows how to do and a cloud-first team would rather skip.
6. Rubrik: S3 inside a security-led Rubrik Security Cloud estate
Rubrik built its name on security-led protection for large estates, and its Amazon S3 protection reached general availability in November 2023.
Pick it if: Rubrik already anchors protection in your data center and S3 should live under the same SLA Domains and console.
Under the hood: Connecting AWS takes a CloudFormation stack plus a cross-account role. SLA Domains then apply policy per account, bucket, or tag, and after an initial full backup Rubrik runs incremental forever.
Backups are written to S3 Standard, Standard-IA, or Glacier in an account and Region of your choosing, with S3 Object Lock as an option. Object search works by name, and Rubrik states that full buckets and single objects can go back to any Region in any account.
What you’ll pay (as of October 2026): There’s no public price list, every edition says Contact Sales, and Enterprise Edition comes as a three-year subscription. Backups stored in your own account show up separately on your AWS bill.
Limits: Rubrik’s 2023 announcement scoped protection to buckets in S3 Standard or S3 Standard-IA, so ask about archived storage classes during the evaluation.
Rubrik’s own guidance splits cloud protection into customer-hosted (your networking, IAM roles, and security posture to manage) and Rubrik Managed, which by design places data beyond your AWS organization and offers fewer configuration options today.
7. MSP360: cheap off-AWS copies of a handful of buckets
Most people know MSP360 for sending server and desktop backups to S3, so the feature that reads from a bucket sits slightly off the main path.
Pick it if: you’re a small IT team or an MSP, and a scheduled copy of a bucket in another cloud, driven from a Windows machine you already own, covers what you need.
Under the hood: MSP360 Backup for Windows does this through its Cloud to Cloud plans. Choose a source cloud account, select buckets, folders, or files, and set a schedule (weekly copies of an S3 bucket into Microsoft Azure, for example).
Each plan retains a fixed number of versions per file, prunes the rest on a schedule, and can AES-encrypt data before upload. Because the copy lands with a different provider, it also covers losing the AWS account outright, which is cheap insurance at these prices.
What you’ll pay (as of October 2026): Ultimate is $179.99 per computer per year, Windows Server $109.99, and plain Windows $29.99, with storage billed on its own. Ask MSP360 which edition carries Cloud to Cloud plans before you buy.
Limits: A Windows agent runs everything, which strains once you’re dealing with many accounts and millions of objects. Retention is tracked per file version, so it works better as a second copy than as a way to roll a whole bucket back.
Where do Backblaze B2, Wasabi, and Cloudflare R2 fit in an S3 backup plan?
They belong on the storage side of the plan, which is a separate job from the software above. Backblaze B2, Wasabi, and Cloudflare R2 are object storage services that speak the S3 API, so any backup tool supporting an S3-compatible target can send copies to them.
Protecting the objects already in your buckets still takes software that reads them, keeps restore points, and brings back the prefix you lost. Moving a bucket to a cheaper provider changes the storage bill and leaves the backup question exactly where it was.
Treat them as candidate destinations for a second copy, and judge the backup software on its restores.
Ask every vendor for one prefix, from last month, in another Region
Before you sign anything, ask each vendor to restore a single prefix from last month into a different Region, live, and then show you what that restore cost.
It sounds small, and it tests most of this comparison at once: point-in-time depth, prefix granularity, cross-Region mechanics, and whether fees turn a routine restore into a budget meeting.
A vendor who can do it in one screen share has done it before, and the number on the bill is the one your finance team will remember.
Common questions
Can I back up S3 without turning on versioning?
Yes, with software that captures objects on its own. AWS Backup for S3 and Clumio’s Backtrack both require S3 Versioning, while the #1 pick above copies objects incrementally and never depends on it.
Where should S3 backups be stored for ransomware protection?
In an immutable, logically air-gapped vault that production credentials can’t open, such as an AWS Backup logically air-gapped vault or a third-party vault outside the source account. Object Lock on the source bucket helps, but it also locks in objects an attacker already encrypted.
Does backing up S3 to another Region double the cost?
It can come close with replication or AWS Backup cross-Region copies, since each keeps a full second copy, and AWS Backup adds $0.02 per GB of transfer (us-east-1, as of October 2026). Software that restores into any Region from one copy skips the replica bucket.
Find Trusted Cardiac Hospitals
Compare heart hospitals by city and services — all in one place.
Explore Hospitals