Running a data-heavy lab session
What you will achieve. By the end of this tutorial you will have set up a course space with a configured application, prepared a dataset, distributed both to a cohort of students, and run through the workflow you would use during a live teaching session.
How long it takes. About 45–60 minutes the first time, depending on the size of your application image and dataset.
What you need before you start. You should have completed Your first course on Nuvolos tutorial - that is, you have an organisation membership, you have created at least one course, and you understand the role of the Master instance. You should also have a sample dataset on your local machine (a CSV is enough) and a clear idea of which application image (for example, JupyterLab with Python, RStudio, or a Stata image) you want students to use.
Step 1 - Create a course space dedicated to the lab
If you reuse a course from Your first course on Nuvolos tutorial , you may end up mixing tutorial leftovers with real lab material. For this tutorial, create a fresh course.
From the Dashboard, select + NEW COURSE. Enter a name (for example, "Data Lab - pilot") and a description. Select + ADD SPACE.
When the application picker appears, do not skip it - choose the application you want students to use during the lab. Enter a name for it and select + INSTALL. Installing the application now means it is part of the Master instance from the start, ready to be distributed to every student.
Why this matters
A data-heavy lab fails the moment one student is on Python 3.9 and the next on Python 3.11. Setting up the application centrally - once, in the Master instance - is the move that eliminates the "it works on my machine" problem entirely. Every student gets the exact same image.
Step 2 - Configure the application
Open the Master instance and start the application you just installed. Install or configure anything the lab needs: extra Python packages, R libraries, connection strings, environment variables, helper scripts. Treat this like a one-time setup of a teaching laptop - but you only have to do it once, here, for everyone.
If the lab needs a GPU or unusual amount of memory, configure that on the application now. See Application resources in Reference for the available options. If your institution restricts GPU usage, your administrator may need to enable it for the space.
Stop the application when you are happy with the configuration. The configuration travels with the application when it is distributed.
Step 3 - Add the dataset and lab notebook to the Master instance
Switch to the Files view in the Master instance. Upload your dataset and a starter notebook (or script) that students will edit during the lab. Keep the layout simple at this stage - flat is fine for a single lab session.
If your dataset is large or you want to use a database-backed approach instead of flat files, see File storage and Data storage in Reference.
Open the application again, run the starter notebook end-to-end, and confirm that everything works in the Master instance. The Master instance is your rehearsal stage - if it does not run cleanly here, it will not run cleanly for any student.
Step 4 - Distribute everything to current and future students
Distribution sends teaching material to all current and future instances in your course. This means you can distribute to both existing and prospective students in one step. The mechanism is documented in detail under Object distribution; the steps below are the practical lab-session workflow.
From the Master instance, make sure you are in the Current state.
On the Files screen, select your dataset and starter notebook and click STAGE. Recently used applications are automatically added to the distribution; review them and remove any you don't want students to receive. Selecting a folder adds the entire file tree recursively, and any sub-folder structure is preserved.
On the sidebar, click the share icon to open the Stage. Review the staged objects and click CONTINUE.
(Optional) Save the staged objects as a named bundle for future reference. This is useful if you plan to run the same lab next term - you can re-distribute the bundle without rebuilding it.
Select the target. By default, you distribute to all students - every existing instance and every future instance.
Decide on notifications. Send a notification email with a custom message so students know new material is available - useful when the lab is the next morning and they need to confirm their environment in advance.
Select the distribution strategy. Overwrite is the default and almost always the correct choice for a fresh lab.
Review the list and click SHARE OBJECTS. Distributing large applications can take a while to complete - start the distribution at least an hour before the session if your image is large.
Checkpoint
If you have a test student account from Your first course on Nuvolos tutorial, sign in as that student now and confirm that the dataset, notebook, and application all appear in the student instance. Run the notebook end-to-end as the student would. This is the single best way to catch problems before a live session.
Step 5 - Run the live session
During the session itself, your role inside Nuvolos is mostly to watch and support. Each student works in their own isolated instance - they cannot break each other's work, and they cannot break the Master instance. You can:
Continue editing material in the Master instance and re-distribute fixes as needed (use Overwrite to push corrections live).
Switch into student instances if a student gets stuck - Space Administrators have full visibility over every instance in the course.
Use the same application as the students in your own instance to demonstrate steps in real time.
If something goes wrong with the material mid-session - a typo in the notebook, a missing file - fix it in the Master instance, stage just the fixed file, and re-distribute. Students who refresh will pick up the fix; students mid-flow will not have their work overwritten if they have already modified the file locally.
Step 6 - Capture the session
After the session, take a snapshot of the Master instance. A snapshot freezes the exact state of files, code, and configuration at this moment, and you can return to it any time. This makes the session reproducible — for next term, for a student who missed it, or for a colleague who wants to see exactly what you ran. Snapshots are explained in How snapshots and states work in Concepts.
You're done
You have run a complete data-heavy lab cycle: configure once, distribute to everyone, watch students work in identical isolated environments, fix and re-distribute on the fly, and capture the result. The same pattern scales from a one-off lab to a full term-length course.
Where to go next
To repeat the same setup for a parallel section, see How-to › Structure your course.
To turn the lab into a graded assignment, see How-to › Assignments, grading, and feedback.
To run the lab on GPU-enabled machines, see How-to › Courses with GPUs.
Last updated
Was this helpful?