Bunny Storage

What Bunny Storage actually is

Bunny Storage is the object storage product of Bunny.net: a storage zone you fill over the HTTP API, FTP/SFTP, a web file manager, or the S3-compatible API, optionally replicated across regions and delivered through Bunny's CDN with no egress charge. It is not trying to be a full S3 replacement — it is trying to be the place your files live so that Bunny's edge network can serve them fast, cheaply, and without a second vendor's bandwidth bill. In our cloud object storage category it is our pick for delivery-centric workloads rather than general-purpose cloud storage.

The model is a storage zone with a primary region and optional replication regions. The published region map spans Johannesburg, Tokyo, Hong Kong, Seattle, Sao Paulo, New York, Miami, Los Angeles, Singapore, Sydney, London, Stockholm, Frankfurt, Madrid, and Prague — fifteen regions, with Mumbai marked as planned. Two tiers structure the purchase. Standard (HDD) costs $0.01/GB per region for up to two regions, $0.005/GB for each additional region, up to nine regions total. Edge (SSD) costs $0.02/GB per region, up to fifteen regions. Both carry a $1 monthly minimum, and both include what matters most here: traffic from Bunny Storage into Bunny CDN is free, API egress is free, and there are no request fees. Bunny's own framing — "stop paying for accessing and managing your files" — is aimed squarely at the hyperscaler bill structure where reading your own data is the expensive part.

Access is through four doors: the dashboard's file browser, FTP/SFTP, a RESTful HTTP API at regional endpoints (for example https://ny.storage.bunnycdn.com), and an S3-compatible API. The S3 layer deserves a careful note, because it is newer than the marketing suggests: the docs describe it as a public preview that must be enabled when the storage zone is created, with credentials drawn from the zone's Access tab, region codes for Frankfurt, New York, Singapore, London, Stockholm, Los Angeles, Johannesburg, and Sydney, and support for both path-style and virtual-hosted-style URLs, presigned URLs signed with AWS4-HMAC-SHA256, and SHA256 checksum validation. rclone works against it out of the box via the generic S3 provider. That is enough compatibility for migration and for most tooling, and we would still characterize it as an emerging surface rather than a finished platform.

Pricing: what you really pay

The arithmetic is simple enough to do on paper. A single-region Standard zone holding 100 GB costs $1/month at the $0.01/GB rate (and at least $1 because of the minimum). Replicating that same 100 GB across three regions lands at $0.025/GB — the two-region price of $0.02/GB plus one additional region at $0.005/GB — so $2.50/month. The Edge SSD tier at $0.02/GB per region doubles that figure per region and extends to fifteen replication regions. Delivery through a Bunny pull zone costs nothing extra on top of the CDN bandwidth you already buy; delivery outside Bunny's CDN is not the design intent.

Compare that with the storage-and-egress split of the hyperscalers and the shape of the savings is clear: a 100 GB library served 500 GB/month to end users costs, on Bunny, the storage fee plus CDN bandwidth at the pull zone's regional rate ($0.01/GB in Europe and North America) — while on S3 the same traffic carries per-request charges and egress at rates several times higher. For content-heavy properties, delivery is the dominant cost, and Bunny prices it at the low end of our benchmarks. The limits you are trading away are on the storage side: no cold or archive tier, no deep lifecycle policies, and replication region count is a line item rather than an architecture.

One practical detail from the docs that the pricing page does not emphasize: the Edge (SSD) tier's primary region is Frankfurt, with replication extending from there, while Standard zones can be born in any of the regions. Teams whose primary audience is Asia or the Americas should model Standard-versus-Edge on latency to their primary region, not just on the per-GB rate.

Performance & infrastructure

Bunny positions storage inside the same network as its CDN: 119 global PoPs, a 250 Tbps+ backbone, and an advertised up-to-5x faster delivery versus legacy object storage, with automatic replication across six continents and automatic region failover. The published availability figure is a 99.99% SLA for the storage service, and the platform SLA page backs the broader infrastructure with a written outage-credit schedule. Uploads reach a regional endpoint and replicate in the background; once a pull zone is attached, files are cached at the edge and served from the closest PoP, with Perma-Cache available for assets that should never expire from the edge.

The delivery numbers we would cite are the platform ones from our Bunny.net review: 24 ms average global latency attributed to CDNPerf data, direct interconnects with 3,000+ ISPs, and NVMe/AMD hardware at the edge. For storage specifically, throughput depends on your primary region and the tier; the SSD tier exists precisely for workloads where HDD seek behavior is the bottleneck. What we can say from the documentation rather than from testing: replication is automatic once configured, files are accessible before replication completes worldwide, and the pull-zone integration turns storage into a CDN origin in a single dropdown.

Reliability is a function of the replication design. A zone with three replication regions can survive losing a region — Bunny describes failover that keeps data online "even when disasters strike, or a region becomes completely unavailable". A single-region zone is, as everywhere, one region's fate at a time. This is where the $0.005/GB per extra region earns its keep: for anything irreplaceable, the second region is the cheapest insurance in the product.

Where it falls short

The S3 compatibility being in public preview is the first caveat: existing S3 SDKs and tools work, presigned URLs and checksums are supported, but the surface is still moving, and workflows that depend on S3 features beyond core object operations (event notifications, lifecycle transitions, storage classes, bucket policies of S3's depth) should verify each one before migration. The database listing for this product once claimed "version control, replication, CORS support" — replication is documented, but object versioning and CORS specifics were not verifiable on the pages we reviewed, so we would not promise them.

The tiering is shallow. There is no archive tier for cold data, so a backup or media-mezzanine workload that S3 Glacier would serve at a fraction of a cent per gigabyte pays full Standard or Edge rates here instead. Analytics and tooling around storage — inventory, access analytics, fine-grained IAM — are thinner than AWS's. And the ecosystem is exactly as wide as Bunny's: if you want Lambda-style processing on upload or a marketplace of storage integrations, you are outside what this product is.

Finally, the free-egress story has a boundary: it is free into Bunny CDN. Serving storage traffic through another CDN or direct public links outside a pull zone is not what the pricing is built around, and multi-cloud architectures that treat storage as a neutral hub will feel that constraint. Support is Bunny's standard 24/7 ticket offering — good for a mid-size vendor, but there is no storage-specific SLA support tier short of enterprise conversations.

Who should pick this

Our pick for media and content platforms, WordPress and static sites, SaaS asset delivery, game and software distribution, and any workload where files exist primarily to be downloaded by users. If your cost profile is dominated by delivery, the free storage-to-CDN path and per-GB pricing make Bunny Storage the cheapest credible answer we track. It is also a fine home for backups of non-critical data where the S3 API (via rclone) is the transport and cost matters more than archive-tier economics.

We would not pick it as a general S3 replacement for data pipelines that depend on S3 eventing, lifecycle automation, or fine-grained IAM, nor as the archive tier for regulatory retention. And if your architecture requires storage to be a neutral, multi-cloud source of truth, the ecosystem boundary is real. For the full platform picture, see our Bunny.net review; for the security add-on that sits in front of these files, see Bunny Shield.

Key takeaways

• Storage from $0.01/GB per region (Standard HDD, up to 9 regions) or $0.02/GB (Edge SSD, up to 15 regions); $1 monthly minimum.
• Free delivery into Bunny CDN, free API egress, and no request fees — the core cost advantage over hyperscaler storage.
• 15 storage regions across six continents with automatic replication and region failover; 99.99% availability SLA.
• S3-compatible API is real (rclone, presigned URLs, checksums) but in public preview and enabled per zone at creation.
• No cold/archive tier and limited storage-side automation — this is delivery-centric storage, not a full S3 substitute.
• Our pick in cloud object storage for content that exists to be served to users.

Sources

• https://bunny.net/storage/
• https://bunny.net/docs/storage/s3
• https://bunny.net/docs/storage/quickstart
• https://bunny.net/pricing/
• https://bunny.net/sla/
• https://subger.com/en/in/service/bunny-cdn