9  Globus

9.1 Overview

Globus is a secure, point-to-point file transfer platform. Data flows directly between endpoints — it never passes through Globus servers. This makes transfers reliable, fast, and encrypted even for very large datasets.

Globus transfer architecture

9.2 Key Concepts

Term Definition
Endpoint A computer or server that can send or receive data via Globus
Collection A named set of files and folders accessible through Globus
Globus Connect Server Infrastructure connecting storage systems (HPC, Google Drive) to Globus
Globus Connect Personal Desktop app that turns your laptop into a Globus endpoint
Globus Plus Enhanced subscription enabling external sharing and multi-endpoint sync

9.3 When to Use Globus

Note

Use Globus for:

  • Datasets larger than 1 GB
  • Transfers between HPC clusters or institutions
  • Sharing data with collaborators at other universities
  • Automated or scheduled transfers

For files under 100 MB, SCP or OnDemand is simpler.

9.4 Getting Started

9.4.1 Create an Account

  1. Go to https://www.globus.org and click Log In
  2. Search for North Carolina State University
  3. Authenticate with your NCSU (Unity ID) credentials
  4. Complete first-time setup prompts

Globus login screen

9.4.2 Connect Your NCSU Google Drive

  1. Go to Collections on the Globus website

    Collections page

  2. Search for ncsu google → click NC State Google Drive Connector

  3. Click Collections → Add Guest Collection

  4. Grant permissions when prompted

  5. Select a directory, enter a display name (e.g., “My Research Data 2025”), and click Create Collection

9.4.3 Connect Your Local Machine

  1. Download Globus Connect Personal from the Collections page (link in top right)

    Globus Connect Personal download

  2. Install and launch; log in with NCSU credentials (browser popup)

  3. Enter a display name for your machine (e.g., “My-Laptop”)

Your machine now appears under Collections.

9.5 Transferring Files

  1. Open File Manager

  2. Set your source collection (left) and destination collection (right)

    File Manager with source and destination

  3. Navigate to and select the files or folders to transfer

  4. Configure Transfer & Timer Options as needed (sync mode, label, email notification)

    Transfer options

  5. Click Start

    Start transfer button

  6. Monitor progress in Activity (left sidebar)

TipTransfer Best Practices
  • Label transfers so you can find them in your Activity history
  • Enable email notifications for large transfers
  • Use sync mode to skip files that haven’t changed
  • Check the Activity log after completion to confirm success

9.6 Scheduled Transfers

  1. Configure a transfer as usual
  2. Click Schedule Start instead of Start
  3. Select date and time; complete setup

Schedule transfer dialog

9.7 Sharing Data with Collaborators

9.7.1 Create a Shared Collection

  1. Navigate to your collection in File Manager

  2. Select the folder to share → click Share

    Share collection button

  3. Add collaborator email addresses or Globus IDs

  4. Set permissions (read-only or read-write)

  5. Set an expiration date if appropriate

  6. Click Add Permission

Warning

Before sharing: verify you have permission to share the data, confirm sensitive data is appropriately protected, and follow your institution’s data management policies.

9.7.2 Access a Shared Collection

  1. You receive an email with the collection name
  2. Go to Collections and search for the name
  3. Authenticate if prompted

9.8 Linking Multiple Institutional Identities

If you’re at multiple institutions, link identities to access endpoints from all of them:

  1. Settings → Accounts → Link Another Identity

    Link identity screen

  2. Authenticate with your other institution

  3. Transfer freely between both organizations’ endpoints

9.9 Advanced: CLI and Python SDK

Globus CLI:

$ pip install globus-cli
$ globus login
$ globus transfer <source_endpoint>:/path/file.txt \
                <dest_endpoint>:/path/file.txt \
                --label "My Transfer"

Python SDK:

from globus_sdk import TransferClient, TransferData

tc = TransferClient()
tdata = TransferData(tc, source_endpoint_id, dest_endpoint_id,
                     label="Automated transfer")
tdata.add_item("/source/path/file.txt", "/dest/path/file.txt")
result = tc.submit_transfer(tdata)
print(f"Transfer ID: {result['task_id']}")

9.10 Troubleshooting

Issue Solution
“Permission Denied” Check permissions on both endpoints; re-authenticate
Transfer very slow Check network; schedule during off-peak hours; bundle small files into archives
Cannot find endpoint Ensure Globus Connect Personal is running; check endpoint visibility
Transfer fails or stalls Check Activity log for errors; verify storage space; cancel and restart
Authentication timeout Re-authenticate by clicking the endpoint; check if institutional credentials expired
Tip

When contacting support, include your Transfer ID (visible in the Activity log) and a screenshot of the error.

9.11 Performance Tips

  • Bundle small files into .tar.gz archives before transferring — many small files transfer much slower than one large archive
  • Sync mode skips files that haven’t changed, saving time on repeat transfers
  • Schedule off-peak to avoid network congestion
  • Filter unnecessary files (caches, temp files) before initiating large transfers

9.12 FAQ

How long do transfers take? Expect 50–100 MB/s on good connections. Globus shows time estimates in the Activity log.

Can I close my browser? Yes. Transfers continue on Globus infrastructure regardless of whether your browser or laptop is on.

Is there a file size limit? Globus itself has none. Individual endpoints may have limits — check with your sysadmin.

What does Globus cost? Basic access is free for all NCSU users. Globus Plus ($50/year) adds sharing and synchronization features.

9.13 Resources