3 Computing Resources at the University of Oregon
After completing this chapter, you will be able to:
- Use your UO digital identity (Duck ID, Duo, UO email, VPN) to reach university services, and know when the VPN is and is not needed
- Find the right place to ask for help at UO, and ask in a way that gets a fast answer
- Describe the Microsoft 365 stack, OneDrive, and UO Dropbox, and know which you are eligible for
- Locate synced cloud folders from the command line and handle the spaces in their paths
- Classify data as green, amber, or red and choose storage and tools accordingly
- Protect research data and back up your work following the 3-2-1 rule
- Install site-licensed software and manage references with Zotero
- Explain what RACS and Talapas offer, how accounts work, and how to reach the cluster through Open OnDemand
The previous chapters equipped your laptop. This chapter is about everything the University of Oregon provides beyond your laptop: the accounts that unlock its services, the places where you can store and share files, the rules that govern what may go where, the software the University licenses on your behalf, and the research computing cluster, Talapas, that you will use later in the course. Much of this is mundane, but it is the kind of mundane that, ignored, costs people a thesis chapter, a data-security incident, or a week of waiting for a cluster account. An hour spent understanding it now is well invested.
3.1 Your UO Digital Identity
Almost everything in this chapter begins with your Duck ID, the UO username and password that you manage at https://duckid.uoregon.edu. It is the key to email, Canvas, Microsoft 365, the software download site, and Talapas. Most UO services also require two-step login through Duo: after entering your password, you approve the login on a second device, usually the Duo Mobile app on your phone. Set Duo up before you need it; being unable to log in to a service you need in a hurry is a frustrating way to discover that your phone is not enrolled.
A Duo prompt is a question: is this you logging in right now? If a push arrives when you are not logging in, deny it—it means someone else has your password and is trying to use it. Attackers count on “MFA fatigue,” sending pushes until a tired user taps approve just to silence the phone. Deny the push and change your Duck ID password immediately.
Your UO email address (you@uoregon.edu) doubles as your sign-in name for Microsoft 365 and for UO’s Dropbox contract. When a service asks you to “sign in with your work or school account,” this is the account it means.
On campus, connect your laptop to the eduroam Wi-Fi network, signing in with your full UO email address and password. The same credentials then work automatically on eduroam networks at thousands of universities and research institutes worldwide—a small perk that pays off at every conference. The “UO Guest” network is for visitors and does not reach campus-restricted resources.
The UO VPN, delivered through the Cisco Secure Client and downloadable from https://uovpn.uoregon.edu, makes your off-campus computer appear to be on the campus network. You only need it for campus-restricted resources such as some library databases, departmental file shares, and some licensed software that must be activated from the campus network. Most UO services, including email, Canvas, Microsoft 365, Teams, and Zoom, work from anywhere without it.
For Talapas, the VPN is recommended but not required. Without it, connect over SSH to login.talapas.uoregon.edu, which is a load balancer that distributes users across the login nodes; expect Duo prompts. With the VPN active, you may also connect directly to an individual login node, login1 through login4. The mechanics of SSH and job submission are in Chapter 20.
3.2 Where to Get Help at UO
There is a lot of help available at UO, and a large part of getting unstuck quickly is knowing which door to knock on. The UO Service Portal at https://service.uoregon.edu is the front door for all campus technology: it has a searchable knowledge base and a ticketing system, and most questions about accounts, software, Wi-Fi, and cloud storage already have an article there, so search it first. Behind the portal sits the Technology Service Desk, reachable by phone at 541-346-4357 or through live chat.
For data analysis rather than technology, UO Libraries Data Services (https://library.uoregon.edu/data-services) is an underused treasure. They run free workshops on R, Python, data analysis, and related topics; offer one-on-one consultations, in person or on Zoom, on statistics, R, Python, version control, and data-management plans; and staff a drop-in help desk in Knight Library on weekdays from 11am to 4pm.
For research computing, RACS (Research Advanced Computing Services), which runs Talapas, can be reached at racs@uoregon.edu or through their service desk at https://hpcrcf.atlassian.net/servicedesk/customer/portal/1; Section 3.10 describes what they do. Closer to home, the Center for Biomedical Data Science (CBDS) at the Knight Campus, directed by the author of this book, supports the computational and data-science side of biomedical research at the Knight Campus, including training such as this course; ask the instructor about CBDS resources and events. Your home department and the Knight Campus also have IT staff who can help with lab computers and shared storage.
How to ask for help well
Whoever you ask, the quality of the answer depends on the quality of the question. Start by saying what you were trying to do (“I’m trying to install R on my M2 Mac”), then include the exact error message as copied text, not a blurry photograph of your screen (a screenshot is acceptable only if the text cannot be copied). Show the command or code you ran and what you expected to happen, say what you have already tried (restarted? searched the error? checked your location with pwd?), and state your system: macOS, Windows, or Ubuntu, and the chip type. For code problems, reduce the problem to a minimal example, the smallest piece of code that still shows the failure. Half the time, the discipline of writing a good question leads you to the answer yourself; the other half, it lets someone else answer in a single reply.
3.3 The Microsoft 365 Stack
Every UO student and employee has a Microsoft 365 license, which you access by signing in at https://office.uoregon.edu with your UO email address. Table 3.1 lists what it includes.
| Tool | What it is for |
|---|---|
| Outlook | UO email and calendar |
| Word, Excel, PowerPoint, OneNote | Desktop, web, and mobile apps; install them on your own devices from office.uoregon.edu |
| Teams | Chat, meetings, and shared files for groups and labs |
| OneDrive | Your personal cloud storage (1 TB) |
| SharePoint / Teams files | Shared storage owned by a group rather than a person |
| Copilot Chat | UO-protected generative AI assistant, when signed in with your UO account |
OneDrive
OneDrive is your personal cloud storage, available at no cost with 1 TB of space. The OneDrive sync app for Mac and Windows creates a OneDrive folder on your computer, and files you save there are copied to the cloud automatically, so they survive a lost or dead laptop. You can share files and folders with UO colleagues and with outside collaborators. OneDrive is approved for FERPA and HIPAA data (see Section 3.4), and individual files may be as large as 250 GB.
OneDrive belongs to you, and when you leave UO it goes away. Data that belongs to a lab, such as raw experimental data, shared analysis code, and protocols, should live in storage the lab controls: a Teams/SharePoint site, a lab Dropbox team folder, or the lab’s project space on Talapas. Otherwise, it leaves when you do.
UO Dropbox
UO also holds an institutional Dropbox contract, but eligibility changed in 2026. Since March 2026, Dropbox accounts under the UO contract are available to faculty, staff, officers of administration, and graduate employees (GEs); students who are not UO employees no longer have access and should use OneDrive instead. If you are eligible, sign in with your UO email address and install the desktop app so that your files sync to a local folder; individual accounts have 2 TB of storage, and labs and departments can purchase team storage. Your lab may also share a Dropbox team folder with you regardless of whether you have your own account. Dropbox is approved for green and amber data, including FERPA records, but not for HIPAA or other red data.
Synced folders have real paths
Both OneDrive and Dropbox create an ordinary folder on your computer, which means it has a path you can use from the command line and from your scripts. Typical locations look like these (yours may differ slightly):
Notice the backslashes: names such as University of Oregon Dropbox contain spaces, and the shell uses spaces to separate arguments, so a path with spaces must be quoted (cd "University of Oregon Dropbox") or escaped with backslashes. Tab completion adds the escapes for you, and Chapter 5 has more on paths and quoting. Find your own synced folder with cd, ls, and Tab completion. Synced folders are excellent for documents and drafts, but large or rapidly changing analysis files are better kept on local disk or on Talapas, where the sync client cannot interfere with them.
3.4 Data Classification: What Can Go Where?
UO classifies data by the harm that would result if it were exposed, and the classification determines where it may be stored and which tools may touch it (University of Oregon, 2019). Know the class of your data before you choose where to put it. Table 3.2 summarizes the three classes; the Information Security Office maintains the authoritative, and much longer, table of which services are approved for which class (University of Oregon Information Security Office, 2024).
| Class | Examples | May be stored in |
|---|---|---|
| Green (low risk) | Published data, your code, most de-identified research data | Almost anywhere, including AI tools |
| Amber (medium risk) | FERPA student records, unpublished research data | OneDrive, Dropbox, Teams/SharePoint |
| Red (high risk) | HIPAA / protected health information, Social Security numbers, credit-card numbers | Only approved systems (for example OneDrive for HIPAA); ask first |
Only green data may be entered into AI tools that UO does not officially support. Microsoft Copilot Chat, signed in with your UO account, is approved for more; see Section 3.9.
Protecting research data
Two categories deserve special attention. Human-subjects data, meaning anything covered by an Institutional Review Board (IRB) protocol, and protected health information covered by HIPAA are red, high-risk data. Store them only where your IRB protocol and UO approve: UO OneDrive and Teams are approved for HIPAA data, Dropbox is not, and personal cloud accounts (a personal Google Drive or iCloud), email attachments, and unsupported AI tools are never acceptable. De-identify data as early in the workflow as you can, and keep the key that links identifiers to participants separately and securely. Encrypt your laptop (FileVault on macOS, BitLocker or Device Encryption on Windows) and use a screen lock, so that a stolen backpack is an inconvenience rather than a breach. Finally, remember that unpublished data and code are valuable; do not post them publicly until your PI agrees. When in doubt, ask before you store or share, whether your PI, the IRB, or the Service Portal. It is far easier than cleaning up afterwards.
3.5 Backups and the 3-2-1 Rule
Every year, someone loses a thesis chapter to a dead laptop or a stolen bag. The standard defense is the 3-2-1 rule: keep three copies of anything important, on two different kinds of storage, with one copy off-site. In practice that might mean the working copy on your laptop, a versioned backup on an external drive using Time Machine (macOS) or File History (Windows), and an off-site copy in OneDrive or your lab’s Dropbox or server. For code, GitHub adds a fourth layer: every commit you push is a backup with full history (Chapter 19).
OneDrive and Dropbox protect you against a dead laptop, but not against yourself. If you delete or overwrite a file, the deletion syncs everywhere within seconds. Keep versioned backups (Time Machine, File History, Git) in addition to sync, and test that you can restore a file before the day you need to.
3.6 Choosing Where to Store Your Work
With several storage options available, it helps to have a rule of thumb. Table 3.3 summarizes what each is good for.
| Storage | Best for | Covered in |
|---|---|---|
| Your laptop’s disk | Active work: fast, but not backed up by itself | Chapter 4 |
| OneDrive | Personal documents, drafts, backups | This chapter |
| Dropbox (if eligible) | Lab-shared folders, collaboration | This chapter |
| Teams / SharePoint | Group-owned documents | This chapter |
| GitHub | Code and small text files, with full version history | Chapter 19 |
| Talapas | Large datasets and heavy computation | Chapter 20 |
The short version: code goes to GitHub (Chapter 19), documents go to OneDrive or Dropbox, and big data goes to Talapas (Section 20.5). Whatever else you do, keep your raw data somewhere backed up and untouched.
3.7 Site-Licensed Software
The University pays for a number of software licenses that you can use for free. The UO Software Downloads site at https://software.uoregon.edu (sign in with your Duck ID) offers MATLAB, Mathematica, LabVIEW, SPSS and SPSS Amos, ESRI ArcGIS, and more; some titles need a license code or a connection to the UO network or VPN to activate, and availability can differ between students and employees, so check the site for the current list. The Microsoft 365 desktop apps are installed from https://office.uoregon.edu. For Adobe products, students get Adobe Express on personal devices, while the full Creative Cloud suite is available in campus computer labs. UO-managed computers (as opposed to your personal laptop) install approved software through the Software Center app.
It is worth noticing that none of the tools at the center of this course, the shell, R, Python, Quarto, LaTeX, Git, VS Code, and Positron, require a license at all. They are free and open source, they will follow you to any future institution or employer, and nothing you build with them will stop working when a contract lapses.
3.8 Managing References with Zotero
You will read hundreds of papers before your qualifying exam, and a reference manager is the only sane way to keep track of them. Zotero (https://www.zotero.org) is free and open source. Its browser connector saves a paper’s citation, and usually the PDF, from a journal page or PubMed with one click; group libraries let you share a reading list with your lab or rotation group; and it inserts citations and bibliographies into Word and into Quarto documents (Insert → Citation in the visual editor). Citing from Quarto is covered in Chapter 16, and a well-kept Zotero library makes it painless. Start now, while the library is small.
3.9 Generative AI at UO
UO provides Microsoft Copilot Chat (https://copilot.microsoft.com) as a UO-protected generative AI assistant; sign in with your UO account. UO-supported AI tools are approved for specific data classifications, while unsupported tools are approved for green data only; the current list of tools and what each may be used with is on UO’s AI Service Offerings page. The course policy follows UO’s guidance for instructors (University of Oregon Teaching Support and Innovation, 2026) and is simple: you may use AI to help you learn, debug, and get suggestions, but the work you submit must be your own, and you say when and how you used an assistant. These tools are excellent at explaining error messages and suggesting code, and confidently wrong surprisingly often, so always read, run, and understand what they produce. We look at AI-assisted coding in depth in Chapter 18.
3.10 Research Computing: RACS and Talapas
Research Advanced Computing Services (RACS), at https://racs.uoregon.edu, runs the University’s research computing and supports UO researchers with computing of every kind. Their best-known service is the Talapas HPC cluster, with CPU, GPU, and large-memory nodes and more than 225 scientific software packages installed. Beyond the cluster, RACS provides large-scale research data storage (roughly $35 per terabyte per year) for data that does not fit in what Talapas includes; Globus for fast, reliable transfer of very large datasets, for example from a sequencing core or a collaborator; consulting on software installation, code optimization, data security, and hardware purchases; grant support for the computing and data-management sections of proposals; and help reaching resources beyond UO, including national supercomputers through NSF ACCESS and cloud consulting for AWS and Azure. Contact them at racs@uoregon.edu or through the RACS service desk. They are there to help; ask early.
Your laptop vs. Talapas
Computation can happen in three kinds of places (Figure 3.1). Your laptop or lab workstation is fast to start and entirely under your control, but limited in memory, storage, and uptime. A cluster such as Talapas is many computers, called nodes, working together and shared among many users through a scheduler that decides whose job runs where. Cloud computing rents machines by the hour from a provider such as Amazon Web Services, Microsoft Azure, or Google Cloud, scaling up on demand and billing for what you use. Chapter 4 compares the three in more depth; for most work in this course the choice is between your laptop and Talapas, and the deciding factor is almost always memory.
To understand why a cluster matters, compare it with the laptop you are reading this on. Table 3.4 puts the numbers side by side.1
| Typical laptop | One standard Talapas node | Talapas large-memory node | All of Talapas | |
|---|---|---|---|---|
| CPU cores | 8-12 | 128 | varies | 14,500+ |
| RAM | 8-16 GB | 512 GB | up to 4 TB | ~129 TB |
| GPUs | integrated | – | – | 89 (NVIDIA A100, H100, V100) |
| Storage | 0.5-1 TB SSD | shared | shared | ~3 PB (3,000 TB) |
A single standard node has roughly ten times the cores and thirty times the memory of a laptop, and Talapas has hundreds of nodes. Your storage there consists of a 250 GB home directory plus your lab’s project space, 2 TB by default (Section 20.5). When an analysis reports “out of memory,” or would take days on your laptop, it is a job for Talapas. As Chapter 4 explains, memory is usually the binding constraint in data analysis, and the large-memory nodes exist precisely for the assemblies, image stacks, and single-cell matrices that will not fit anywhere else.
Getting an account
Every Talapas user is sponsored by a UO principal investigator and belongs to at least one PIRG (Principal Investigator Research Group), which the PI is responsible for. Ask your PI, or your rotation PI, to add you, and request the account through RACS’s request page at https://racs.uoregon.edu/request-access (which leads to the Talapas customer portal, https://hpcrcf.atlassian.net/servicedesk/customer/portal/1/group/1/create/9). A PI who does not yet have a PIRG can request one from racs@uoregon.edu. Setup can take a few days, so ask early; for this course, class access will be arranged before the Talapas material in Chapter 20. The Talapas Knowledge Base is the authoritative documentation.
Open OnDemand: Talapas in your browser
You do not need to master SSH (Section 20.4.1) to start using Talapas. Open OnDemand at https://ondemand.talapas.uoregon.edu gives you the cluster through a web browser; log in with your Duck ID using Chrome or Firefox (a private window avoids login glitches). From there you can launch Jupyter Lab notebooks running on Talapas; a full remote desktop with RStudio Server, MATLAB, and Stata; a shell in the browser; a file explorer for drag-and-drop transfer between your laptop and the cluster; and a job composer for submitting batch jobs. Because OnDemand apps run as jobs on compute nodes, they keep running even after you close your browser, so remember to end sessions you no longer need. Command-line access over SSH and job scheduling with SLURM are covered in Chapter 20.
3.11 Summary
This chapter mapped the University’s computing landscape so that you can find your way around it. Key takeaways:
- Your Duck ID with Duo unlocks nearly everything; your UO email is your Microsoft 365 and Dropbox sign-in; the VPN is needed only for campus-restricted resources and is recommended but not required for Talapas
- Help is available from the Service Portal and Technology Service Desk, UO Libraries Data Services (workshops, consultations, drop-in desk), RACS, and CBDS; a good question includes the exact error, the command, your system, and what you tried
- Everyone has Microsoft 365 with 1 TB of OneDrive; UO Dropbox is limited to faculty, staff, and GEs; synced folders have real paths, usually with spaces that must be quoted
- Data is green, amber, or red; IRB and HIPAA data are red; only green data may go into unsupported AI tools; encrypt your laptop and de-identify early
- Follow the 3-2-1 backup rule and remember that a synced folder is not a backup
- Code goes to GitHub, documents to OneDrive/Dropbox, big data to Talapas
- UO licenses MATLAB, Mathematica, and more at software.uoregon.edu, but the core tools of this course are free and open source
- Zotero manages references; Copilot Chat is UO’s protected AI assistant
- RACS runs Talapas (14,500+ cores, ~129 TB RAM, 89 GPUs, ~3 PB storage), provides storage, Globus, consulting, and grant support; every user belongs to a PIRG; Open OnDemand offers Jupyter, RStudio, a shell, and file transfer from a browser
Next, Chapter 5 opens the terminal and begins the command-line skills that run through the rest of the book.
Exercises
Find your synced folder: Sign in to OneDrive (and Dropbox, if you are a GE) with your UO account and install the sync app. From a terminal,
cdinto the synced folder, runpwd, and write down the absolute path. How did you handle the spaces in the name?Classify three datasets: For each of the following, state whether it is green, amber, or red data and where you could store it: (a) a table of Young’s modulus measurements from a hydrogel you made; (b) survey responses from human participants under an IRB protocol, including names; (c) the R scripts that made the figures for your last paper.
Set up a backup: Configure Time Machine or File History with an external drive, or verify that your OneDrive folder is syncing. Then delete a test file from a synced folder and try to recover it. Did the recovery work? Would it have worked for a file you overwrote three weeks ago?
Make the 3-2-1 plan: For the most important document you own (a thesis draft, a manuscript, a dataset), list where the three copies are or will be, which two kinds of storage they use, and which copy is off-site.
Start your Talapas account: Ask your PI whether their lab has a PIRG and, if so, to add you. Then log in to Open OnDemand and open the file explorer to look at your home directory. What is the path?
Cluster hardware changes as nodes are added and retired. The figures here are from the RACS documentation in September 2026; the current numbers are always at https://racs.uoregon.edu.↩︎