The Collaboratory Bucket service was initially called Data Proxy. The old name may still appear in some of our documentation.

Introduction

Collabs offer their users two forms of storage: the Drive and the Bucket. They each come with their respective advantages. The Bucket vs Drive section below covers the main differences.

Bucket storage is implemented as object storage using Openstack Swift technology. Data in this type of storage is organized in containers which are also often called buckets to avoid the word container which is used in many different contexts to refer to unrelated concepts.

Just as for the Drive, files stored in the Bucket of a collab all have the same collab-wide permissions. Some users of the collab Team have read access, others have write access, but there are no specific permissions to individual files or folders.

The bucket of a public collab is not visible to visitors without an EBRAINS account nor to users that are not a member of the Team of the collab.

Buckets support a public link feature. An editor or admin of a collab can create a public link to a file in a bucket of the collab to share the file beyond the members of the collab.

The main  use cases for the Bucket include:

• datasets uploaded by EBRAINS users
• videos
• large IT images, although EBRAINS has a self-hosted Docker registry which is better suited for that specific use case than the Bucket service

Users can access these Bucket files programmatically (e.g. via Jupyter Notebooks) by using the bucket API.

Bucket vs Drive

Advantages of the Bucket

• The Bucket can store much larger files. We will be setting a file size limit in the Drive.
• The Bucket has much greater network bandwidth, supporting also video streaming.
• Users get larger quotas for storage in Buckets.

Advantages of the Drive

• Files in the Drive can be moved and renamed without having to download and re-upload them.
• Versions of a file are readily accessible to users in the Drive. 
• Office file types can be opened for online collaborative editing with the Collaboratory Office service directly from the Drive.
• Access from a Jupyter notebook can be performed without importing any library/package.