Datahub and DSMLP Overview¶
This page describes Datahub and DSMLP and names the starting page for each audience. It also summarizes standard features, platform constraints, support contacts, and policy.
Datahub, DSMLP, and the Research Cluster¶
UC San Diego's Datahub and Data Science & Machine Learning Platform (DSMLP) provide thousands of undergraduates, graduate students, and their instructors with access to Jupyter, RStudio, and other computational resources for scheduled classes, formal independent study, and student projects. The same cluster runs research work and, in that context, is also called the Research Cluster.
The platforms are used in courses ranging from introductory Data Science lectures to graduate-level applied machine learning, and in curricula in Biology, Music, Social Sciences, and Public Health. In AY 2025-26, the platforms hosted more than 140 classes, 75 instructors, and 18,000 student enrollments.
Datahub is the web interface, at datahub.ucsd.edu. DSMLP is the cluster that runs Datahub sessions, and it is also reachable from a terminal. Datahub and DSMLP are a single service, and the same files are present in both.
Where to Begin¶
Each audience has its own starting page.
| Audience | Starting page | Assumes |
|---|---|---|
| A student enrolled in a course that uses Datahub | Using Datahub in a Course | A web browser and nothing else |
A student in such a course, where the course requires SSH, launch.sh, batch jobs, VS Code, or a GPU class the course menu does not offer |
Working from the Command Line | A terminal and ssh |
| Teaching or assisting with a course | Teaching with Datahub and DSMLP | A browser for most tasks |
| Undertaking a personal project, independent study, capstone, or club or team project | Projects & Independent Study | Familiarity with a shell |
| A graduate student, postdoc, undergraduate, or staff researcher working without a lab workspace | Research on DSMLP | Familiarity with a shell |
| Faculty arranging access for a research group | Setting Up a Research Lab | Some steps are performed by ITS staff |
Coursework and a personal project are two separate workspaces, with different eligibility, different resources, and different access lifetimes. Coursework is covered in Using Datahub in a Course and Working from the Command Line, and a personal project in Projects & Independent Study.
Workspaces¶
Users are divided into groups called workspaces: one per course, one per lab, or one general catch-all population. A workspace anchors all cluster configuration: rosters, storage, container images, GPU access, and storage and GPU quotas, as described in What a Workspace Is and What It Controls.
A member may belong to several workspaces at once, for example as a TA for one course, a student in another, and a member of a lab, and chooses the workspace to work in at launch. A member of more than one workspace has a home directory in each, as described in Belonging to Several Workspaces.
See also: Workspaces & Storage
Standard Features¶
Job Environment¶
All jobs, whether browser-based Jupyter sessions, command-line jobs, or batch jobs, run in a compute environment derived from the member's account and workspace together with any job-specific configuration. A course session in the browser typically begins at 2 CPU cores and 4GB RAM, as described in The Browser Session. Upper limits are based on class size, demand, and capacity.
GPUs¶
GPUs are organized into five classes by size rather than by hardware model. Each workspace is given access to one or more classes, matching its anticipated work. The classes are listed in GPU Class Sizes.
Storage¶
Each member has a per-user, per-workspace home directory; a shared area readable by everyone in the workspace; and a per-user, cluster-wide private area available in every workspace. The directories are listed in Where Files Live, and the quotas for course and research home directories in Workspace and Personal Quotas. Optional mounts of external storage are available, as described in Mounting External Storage.
Standard Software Images¶
Curated images cover the widely used Python, R, and Julia data analysis stacks. They include a CUDA/GPU-enabled image with TensorFlow and PyTorch, and an RStudio image. Courses and labs may further customize or pin their own images. The images are described in Standard Images, Tags, and Pinning.
Ways to Work¶
Work is done in a browser, at a terminal over SSH, or in a desktop editor such as VS Code connected to a cluster container. Each route is described in Access.
GPU Reservations¶
From Fall 2026, access to GPUs is managed by a reservation system. A booking, made in the reservation app at reserve.dsmlp.ucsd.edu, holds capacity for a specific window, and for the length of that window sessions are admitted ahead of the walk-up queue. Ad-hoc, on-demand use without a booking is also supported. Booking is described in Reservations. Reservations, per-group quotas, and Service Unit budgets are the mechanisms that manage contention for GPUs, as described in GPU Access.
Warning
An on-demand launch draws on the Service Unit budget as a booked window does, and launching an eligible session authorizes that spend. See On-Demand Lease Charges.
Platform Constraints¶
Root Access¶
Containers run unprivileged, under the member's own UID, with no root or sudo
inside the container, as described in
Root Access and System Packages.
sudo apt-get install fails by design, and the failure is not a fault to
report.
Computation on the Login Node¶
The login node is for launching jobs and moving files, and running computation on it is prohibited, as described in What the Login Node Is For.
Idle GPU Sessions¶
An idle GPU session still holds its GPU, and the session is reclaimed once it has been idle long enough, as described in What Counts as Idle.
Support¶
Questions and feedback go to the contact for the topic.
| Topic | Contact |
|---|---|
| A course a student is enrolled in | The instructor or TA, first |
| Datahub or DSMLP itself | datahub@ucsd.edu, or the ITS Service Desk |
| The Research Cluster, or Universal Scale Storage | rcd-support@ucsd.edu |
| Which platform is appropriate for a research project | Research IT |
| 1:1 Consultation, for instructors, TAs, and Technical Points of Contact (TPOCs) | ucsd-datahub.youcanbook.me |
Urgent or broadly scoped problems may be escalated through the ITS Service Desk. A ticket for such a problem states what is affected and how many people, as described in Incidents Affecting More Than One Person. ITS response targets are given in Response Targets, the terms of 1:1 Consultation in Support & Technical Consultation, and all support routes in Getting Help.
Policy¶
Conditions of use are set out in Policy.
- Highly sensitive data (P4), such as clinical records or export-controlled information, must not be used on Datahub or DSMLP, and legally or contractually protected information (P3) may be permitted after review, as described in Data Classification.
- Datahub and DSMLP system resources are shared among all courses and research groups, and demand, in particular for GPUs, may exceed capacity at peak hours during 10th and Finals Weeks or at assignment deadlines, as described in Use of Shared Resources.
- Datahub may be unavailable during scheduled maintenance for time-sensitive updates or security patches, on the schedule given in Scheduled Maintenance.
- Datahub and DSMLP are not to be used to host externally available services or applications except as required for coursework or projects, as described in Hosting Externally Available Services and Availability and Reliability.
- The campus-wide IT policy on acceptable use applies to Datahub and DSMLP, as described in Appropriate Use.