Backup Data vs Backblaze B2: a backup system or a cheap bucket?

Let's be clear up front, because it decides the whole comparison: these aren't the same kind of thing.

• Backblaze B2 is object storage. It is arguably the best-value S3-compatible bucket you can buy, with famously transparent pricing and generous egress terms. What it stores is bytes. It has no idea what a snapshot, a retention policy, a deduplicated chunk, or a verified restore is.
• Backup Data is a backup system. Storage is included, but the product is the logic around it: content-addressed dedupe, point-in-time snapshots, retention pruning, client-side encryption, and restores that verify integrity, all driven by a Go/JS SDK.

So the real question isn't "which is cheaper per gigabyte." B2 wins that outright. The question is: do you want to buy storage and build the backup system yourself, or buy the backup system?

If you already run a backup tool (restic, Duplicati, Duplicacy) and just need somewhere cheap to put the data, B2 is an excellent answer and this page will tell you to use it. If "our backup strategy" is a shell script that pipes a dump into a B2 bucket, read on, because that setup is missing most of what makes a backup trustworthy.

A quick disambiguation: Backblaze sells two different things. B2 Cloud Storage is the developer object store this page compares against. Backblaze Computer Backup is a separate consumer product that backs up laptops and desktops for a flat per-machine price. It's a fine product, but it doesn't back up servers, databases, or CI, so it isn't the comparison for this audience.

Head to head

The differences that actually matter

1. A bucket stores bytes; it doesn't protect them

This is the whole thing. When you aws s3 cp (or b2 upload-file) a dump into B2, you get a copy of a file in a durable, cheap place. That is genuinely valuable, and B2's durability is not in question. But a backup is more than a stored file:

• Which of these 40 uploaded dumps is internally consistent and actually restorable?
• When did anyone last restore one and confirm it worked?
• What deletes the old ones on a schedule so storage doesn't grow forever?
• What verifies the bytes haven't rotted before you rely on them?

B2 answers none of those, by design. It's storage. BaaS is the layer that answers all of them, and stores the result. The pg_dump | b2 upload-file script in a lot of repos is the bucket half of a backup system with the other half left as a TODO.

2. "Cheaper per GB" is not "cheaper per backup"

B2 at ~$6/TB-month is cheaper raw storage than BaaS. That comparison is real, and if you store a fixed, deduplicated dataset, B2 wins on price.

But most hand-rolled scripts don't deduplicate. They upload a full dump every night. Thirty nightly dumps of a 20 GB database is 600 GB on B2, and you pay for all 600 GB, because B2 stores exactly what you send it. BaaS uploads the first dump and then only the changed chunks, so thirty retained snapshots of a slowly-changing database cost a small fraction of thirty full copies. The sticker price per GB is lower on B2; the bill for "keep 30 days of database backups" can easily land the other way once dedupe is in the picture. Run your own numbers on the cost calculator rather than trusting either vendor's framing.

3. Who holds the keys

B2 encrypts at rest server-side, and Backblaze holds those keys. You can encrypt client-side before upload, but then you're back to running that tooling yourself (another reason people reach for restic on top of B2).

BaaS encrypts client-side as a built-in option: AES-256-GCM on your machine, the keyfile stays with you, Backup Data stores ciphertext it cannot read. For a "customer-managed keys" requirement, that's built in rather than assembled. The tradeoff is real and worth stating: lose the keyfile and passphrase and the data is unrecoverable.

4. Restore is a product feature, not your weekend

On B2, restore is whatever you build: list the objects, download them, reassemble, decompress, hope the integrity is intact. There's no drill, no verification step, no measured recovery time unless you write it.

BaaS treats restore as the point of the whole exercise: client.restore() verifies every chunk against its checksum before reassembly, and the restore-drill pattern (restore into scratch, confirm, measure RTO) is part of the documented workflow. A backup you can't confidently restore is just expensive hope, cheap per gigabyte or not.

When Backblaze B2 is the right call

Genuinely, pick B2 if:

• You already run a real backup tool (restic, Duplicacy, Duplicati, Kopia) and want the cheapest, most reliable S3-compatible backend to point it at. This is B2 at its best, and it's a great choice.
• You want general-purpose object storage, not only backup: static assets, data lakes, media, archives. B2 is a Swiss-army bucket; BaaS is backup-specific.
• Egress economics dominate your workload. B2's free-egress allowance (up to 3x stored data per month) is genuinely generous and hard to beat for read-heavy or large-restore patterns.
• Budget is the binding constraint and you have the ops capacity to own the backup logic yourself.

When Backup Data is the right call

• You want a backup system, not storage plus glue. Dump, one SDK call, and dedupe, retention, encryption, and verified restore are handled.
• You retain many near-identical backups (nightly DB dumps, checkpoints), where B2's per-byte pricing on undeduplicated full copies adds up faster than a flat tier.
• Client-held encryption keys are a policy requirement, not a project.
• Restore confidence matters: you want checksum-verified restores and a drill you can schedule, not custom download scripts.
• Backups belong to app developers via an SDK in the codebase, not to whoever owns the storage account.

Or: use both

These compose cleanly, and for some teams the best answer uses each for what it's good at:

1. B2 as a cheap archive tier for large, already-packaged artifacts you rarely touch: media, raw exports, cold data.
2. BaaS for the operational backups that need logic: production database dumps and model checkpoints, where dedupe, verified restore, and client-side keys earn their place.

And if you love B2's price and want the backup logic too, the honest DIY path is restic pointed at a B2 bucket, which we compare in BaaS vs restic. That page is the right read if your real question is "managed backup vs a tool I run on cheap storage."

FAQ

Is BaaS just a wrapper around Backblaze B2? No. BaaS stores backup data on Backup Data's own storage network, independent of B2, and adds the dedupe, snapshot, encryption, and restore layer. It isn't reselling a B2 bucket.

B2 is cheaper per GB, so why would BaaS cost less? It often doesn't on raw storage, and we won't pretend otherwise. Where BaaS wins on total cost is deduplication: B2 bills every byte you upload, so undeduplicated nightly full dumps get expensive, while BaaS stores only changed chunks. Compare your actual retention pattern on the cost calculator.

Can I use B2 as the storage backend for BaaS? No. BaaS includes storage as part of the service. If choosing your own backend (B2, S3, SFTP) is a hard requirement, you want a self-run tool like restic, not a managed service.

How is this different from Backblaze Computer Backup? Computer Backup is Backblaze's consumer product for laptops and desktops at a flat per-machine price. It doesn't back up servers, databases, or CI pipelines. For those, the relevant Backblaze product is B2, which is what this page compares.

Does BaaS charge for restores like egress? Restores don't carry a separate egress line item beyond your tier. B2's egress is cheap and partly free, but it is a separate meter you should model if you restore large datasets often.

Try the free tier: backupdata.io offers 5 GB, no card, first verified snapshot in ~10 minutes via the quickstart.