Skip to content

Transfer files with Globus

When to use this

Use Globus when you need to:

  • transfer large datasets (hundreds of GB to TB+)
  • move data between systems (e.g. HPC, storage services)
  • run reliable, unattended transfers
  • ensure data integrity during transfer

If your transfer is small or primarily human-driven, consider Nextcloud instead.

Before you start

Make sure you have:

  • a UCT account
  • access to the source data (e.g. local machine, HPC, or storage system)
  • access to the destination system
  • both systems available as Globus endpoints (or the ability to create one)

Log in to Globus

  1. Go to https://app.globus.org
  2. Select your institution login.
  3. Authenticate using your UCT credentials.

Identify your endpoints

An endpoint is a system that can send or receive data.

Typical endpoints include: - HPC systems - institutional storage - your local machine (via Globus Connect Personal)

To use your local machine

  1. Install Globus Connect Personal.
  2. Set up a local endpoint.
  3. Ensure it is running during transfers.

Transfer data between endpoints

  1. In the Globus web interface, open the File Manager.
  2. Select your source endpoint.
  3. Navigate to the source folder.
  4. Select your destination endpoint.
  5. Navigate to the destination folder.
  6. Select the files or folders to transfer.
  7. Click Start.

Globus will: - queue the transfer - retry automatically if needed - verify file integrity

You do not need to keep your browser open once the transfer has started.

Monitor transfer progress

  1. Go to the Activity or Transfers section.
  2. View status:
  3. active
  4. completed
  5. failed

If a transfer fails, Globus will typically retry automatically.

Good practice

  • Transfer directories rather than many individual files where possible.
  • Avoid modifying files while they are being transferred.
  • Use clear folder structures before starting transfers.
  • Confirm sufficient storage space at the destination.
  • For HPC workflows, transfer data to the correct storage location (e.g. RDS or scratch as appropriate).

When to use something else

Use another method if:

  • the transfer is small and ad hoc → use Nextcloud
  • the workflow is simple and local → use scp or rsync
  • you need browser-based sharing with non-technical users → use Nextcloud
  • services/data-transfer/index.md
  • how-to/data-transfer/transfer-with-nextcloud.md
  • reference/storage/rds.md
UCT eResearch Centre · Research computing, software, data, and infrastructure support