Object Storage¶

Verda Object Storage is S3-compatible storage for large amounts of unstructured data: datasets, model checkpoints, training artifacts, backups, and media. You manage buckets and access keys from the Verda console and the Verda API, and you read and write objects with any S3-compatible tool or SDK.

Ready to store your first object? Jump to the Quickstart.

How it works¶

Object Storage exposes the standard S3 API, so you keep using the tools you already know: the AWS CLI, boto3, rclone, s5cmd, or any S3 SDK. Buckets and access keys are managed in the Verda console and the Verda API, alongside the rest of your project.

At launch, Object Storage is available in one location (FIN-03). Point your S3 client at the S3-compatible endpoint:

Bucket names are globally unique and follow standard S3 naming rules (3 to 63 characters).

Features¶

Manage Object Storage in the console¶

Reading and writing objects happens over the S3 endpoint. Everything around it is managed in the Verda console.

Buckets. Open your project and go to Object storage. The table lists every bucket with its location, size, approximate object count, current cost, and creation date. Create bucket asks for a Bucket name and a Location. Each row offers View metadata (bucket ID, current size, and file count), View CLI commands (ready-made commands to configure the CLI, list buckets, and upload a file), and Delete bucket, which cannot be undone.

Access keys. Go to Project management -> Credentials and use the Object Storage Access Keys section to create or delete keys. The secret key is shown once, when you create it. All project members see every key in the project, along with who created it.

Usage and cost. The Object storage page shows your current usage, the price per GiB, and what your buckets cost so far this month. It also shows the Beta quota that applies to you, and it is the live source of truth for those numbers.

Performance¶

Object Storage handles both large sequential transfers and workloads made up of many small objects.

You get the best throughput when you:

• Run transfers from Verda instances or Instant Clusters in the same region (FIN-03). Traffic stays inside the Finland region instead of crossing the public internet.
• Transfer large files as parallel parts. The Verda CLI uploads and downloads large objects as multiple concurrent parts by default and resumes them if they are interrupted. The AWS CLI, boto3, and other S3 SDKs can do the same through their multipart settings.
• Parallelize requests when you have many small objects, rather than moving them one at a time.

Security and access¶

• Buckets are private by default. Bucket policies, ACLs, and CORS configuration are not supported, so objects are never exposed anonymously.
• Share individual objects with presigned URLs that expire after a period you choose.
• Access is controlled with S3 access keys scoped to your project. Every key in a project can access every bucket in that project. Create and delete keys in the console under Project management -> Credentials, or through the API. The secret key is shown only once when you create it, so store it somewhere safe and treat it like a password.
• Buckets are created in the Verda console or through the Verda API. Creating buckets through the S3 API (for example aws s3 mb or CreateBucket) is not yet supported.

Pricing and billing¶

You pay for the amount of data you store, per GiB per month, charged from your prepaid balance every 10 minutes, like your other Verda resources. Uploads, downloads, and API requests are not charged. There is no subscription and no minimum commitment. The Object storage page in the console shows the current price.

If your balance runs out, S3 access to all buckets in your projects is suspended until you top up. Access resumes automatically after the top-up. Top up before your balance runs out to keep uninterrupted access to your data.

Limits and quota¶

While Object Storage is in Beta, the following default quota applies across all your projects:

To see the quota that applies to you, open Account settings -> Quotas in the console. Trust that number if it differs from the table above. Quotas will rise as Beta progresses. If you need more sooner, request a quota increase from the same Quotas page.

Next steps¶

• Quickstart: create an access key, connect a client, and move your first object.
• Verda CLI: Object Storage: the Verda CLI verda object-storage command reference.
• Verda API: the /v1/object-storage endpoints for buckets and access keys.