Object storage without the catch.
Flat pricing, no lock-in, no surprise egress bills, and a stack you can audit. S3-compatible, so point any app at it: rclone, Obsidian, PeerTube. 100% Canadian Hosted.
Phase 1 Single Montreal node serving alpha.
Canadian infrastructure.
Hosted in Montreal on Canadian-owned hardware. No US cloud. Your bytes don't leave the country.
Open source. Inspect the stack.
Storm Buckets runs on Garage and Storm Pulse. We contribute back upstream to the projects our stack depends on.
What people run on Storm Buckets.
Three copies of your data, on two different machines, one of them offsite. Storm is the offsite one. A dead server, a ransomware run, or a fire takes every copy sitting in your building at once; the copy in Montreal is the one still there afterward. Storm's Backup tool connects to the S3 provider you already use and pulls a copy in on a schedule you set, incrementally, with a failure alert when a run breaks.
The S3 backend an application writes to: user uploads, generated files, exports, media. Point it at the endpoint with an access key and it works through the SDK it already uses, no code change. Connecting S3 tooling →
Deploy a static website straight from a bucket, with no web server in front of it. Hosting is included in the storage block. Storm Sites →
PeerTube, Mastodon, Obsidian sync, anything that speaks S3. The application writes over S3 with its own access key and keeps its media off its own disk. The PeerTube walkthrough →
If your workload looks like one of these, this is the right product.
The copy that survives a fire, a ransomware run, or a dead server sits in Montreal, pulled on a schedule you set, under Canadian law. This is what most of the alpha runs today. Backup guide →
Its objects sit in Montreal rather than making the trip to a US region and back, and the jurisdiction question is settled by where the disks physically are. Point your existing S3 SDK at the endpoint; nothing in your code changes.
The engine is Garage, open source and self-hostable. If Storm ever stops being the right answer, the same stack runs on hardware you own, and your data leaves over the same S3 API it arrived on.
One block size. One invoice you can forecast.
Billing off Billing is not switched on yet. We're focused on building, testing, and hardening the platform. The prices below are what you'll pay when billing goes live. Nothing is charged during the alpha.
A full terabyte. Use what you need inside it.
- ✓ 1 TB block, same bill at 100 GB or 950 GB
- ✓ Stack blocks as you grow
- ✓ Buckets and keys are not billed units
- ✓ Custom domains + static hosting
- ✓ Audit logs
One rung, one price. A terabyte is $9 CAD whether you store 40 GB in it or 950 GB, and the invoice is the same number every month.
Founding Alpha A 90 GB grant for alpha testing, plus all features unlocked. The grant persists past alpha as your permanent Founding Alpha allocation. Create a Storm account, then request Buckets access: Storm activates it per account.
Hosting a static site? The same bucket can serve it, hosting is included in the block.
Need your own private cluster instead of the shared one? Storm Buckets Private →
What's different, and what it costs you.
The big S3-compatible engines are US companies billing in USD. Storm is Canadian hardware under Canadian law, billed in CAD, so the exchange rate never moves your invoice.
The storage engine is Garage and the management agent is Storm Pulse, both open source and public, with our fixes sent back upstream. None of the alternatives ship an open engine. Nothing in the path is a black box you have to take on trust.
Storm's egress allowance is 3x your capacity, and past it reads throttle instead of billing. The invoice is the same number every month by construction, not by you watching a meter.
Price floor Storm is not the cheapest per terabyte and will not be. Canadian-owned hardware run in discrete units by a named operator costs what it costs. The cheaper engines are billed in USD, so the exchange rate moves the real number.
One region One node, in Montreal, and no global edge network. That is a hyperscaler capital expense, and Storm puts the money into Canadian ground instead. A second Canadian zone is Phase 2; a worldwide edge is not on the roadmap.
Egress ceiling The allowance is 3x your capacity per month, and past it reads slow down. That throttle is the mechanism keeping the invoice flat, which makes Storm the wrong pick for a workload whose whole job is high sustained delivery.
Breadth One product, not eighty. No managed database, no queue, no serverless runtime to wire this into. Buckets is learnable in an afternoon, and Storm is not where you consolidate an entire platform.
01 Is Storm Buckets actually S3-compatible?
Yes. Storage runs on Garage, an open-source S3-compatible engine. Standard S3 tools and SDKs work against it unchanged: the AWS CLI, rclone, boto3, s3cmd. Garage publishes its S3 compatibility matrix, so you can confirm coverage before you commit rather than after you migrate.
02 Where does my data physically live?
Montreal, on Canadian-owned hardware. The control plane is Canadian too, which is the part most residency claims skip: a Canadian bucket behind a US control plane is not fully Canadian. Your files stay under Law 25, with no CLOUD Act data-access exposure.
03 How does egress actually work?
Your monthly egress allowance is 3x your total capacity. Cross it and reads are throttled, never billed. There is no metered data-transfer line on the invoice. A throttle is something you notice and recover from; a metered overage is an invoice you find out about after it already happened.
04 How durable is it, honestly?
The Standard tier is single-zone, RAID6-backed, RF1. In plain terms: your data sits on one Montreal node whose disks are RAID6, so it survives drive failures inside that node, and it is not yet copied to a second site. Treat Storm as the offsite leg of a 3-2-1 backup rather than your only copy, and match the tier to how much the data would hurt to lose. Multi-zone replication is Phase 2.
05 What happens if Storm disappears?
It's standard object storage over a standard S3 API. Your tooling already knows how to walk away: point rclone or the AWS CLI somewhere else and copy your data out. There's no proprietary format and no exit fee. Leaving is a copy command.