Teaching with Datahub & DSMLP: Scope of Support & Guidelines for Usage¶

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 advanced computational resources for scheduled classes, formal independent study, and student projects.

The platforms are used in courses ranging from introductory Data Science lectures, to graduate-level applied machine-learning, to curricula in Biology, Music, Social Sciences, and Public Health. Over AY 2025-26, more than 140 classes, 75 instructors, and 18,000 student enrollments were hosted.

This article describes the instructional request process and timeline; the curated Standard Software Images and how to customize them; options for support and technical consulting; how GPU access is managed for a course from Fall 2026; and finally, a few pertinent caveats and conditions including important limitations on use of sensitive/protected data.

One article for both instructors and TAs. Where a task is restricted to one role, it is noted. Most of what follows is done in a browser; customization requires a shell and someone prepared to use it.

Please reach out to our team with any questions or feedback regarding these guidelines or Datahub/DSMLP as a whole.

• Email: datahub@ucsd.edu
• 1:1 Consultation: https://ucsd-datahub.youcanbook.me/

Course Timeline: Requesting Access & Important Dates¶

Instructors, TAs, and departmental staff may request Datahub/DSMLP access via the Specialized Instructional Computing Course Request form beginning the 2nd week of the previous term. (For Summer/Fall classes, submission opens the 2nd week of Spring.)

Requests must be submitted 4 weeks prior to the start of instruction; this includes designation of a Technical Point of Contact when required (see below). Late requests are given lower priority and will be reviewed as time permits. Setup and ultimate availability to students will depend upon individual complexity and overall request load; students may not gain access until as late as 4th week.

Course setup and instructor access opens 4-5 weeks prior to the start of instruction. The instructor, TAs, and other course staff should begin testing features, validating assignments, and making desired customizations. Earlier setup is available upon request. Note that 1:1 Consulting support for customization (discussed below) is limited in the final weeks of each term.

Student access follows TSS course rosters — the system formerly called TritonLink. Rosters are loaded into workspaces one business day before the start of the term, and a subsequent add/drop is reflected by 10am the day following the change. Non-roster auditor and observer access is managed through Canvas; Concurrent Enrollment students receive access by way of Extended Studies staff.

Instructors and students retain access for one additional quarter beyond the instructional term(s), ignoring Summer for Spring courses. Instructors may request individual accounts remain active longer to facilitate Incomplete grade resolution, academic integrity proceedings, course development or hand-off, or similar circumstances.

Course environments are purged one quarter after account deactivation, i.e. after two quarters excluding Summer, except for: individual accounts extended as above; and instructor/course-wide files which can be archived for up to 3 years upon request. Please contact ITS to revive a previously archived class environment, or to make an archive available for download. (Note: large datasets cannot be archived due to storage limitations; contact ITS to discuss alternate options.) → When Access Starts & Ends

For-credit Independent study, capstones, and similar student projects may request ongoing access via the Independent Study Request form — please direct those students to Projects & Independent Study. Research IT can help connect faculty, staff, and student researchers with compute platforms for non-credit research.

Standard Features & Standard Software Images¶

Datahub offers a set of curated software environments which are sufficient for many courses and use cases. Questions, configuration assistance, malfunctions, and errors relating to these standard features receive priority support from IT Services via a number of channels (see below):

• Standard Software Images accessible via web-based Jupyter notebooks as well as text/CLI:
• datascience-notebook includes widely-used libraries for data analysis from the Julia, Python, and R communities.
• scipy-ml-notebook supplements 'datascience-notebook' with CUDA/GPU enabled tools such as TensorFlow and PyTorch for deep learning applications.

• rstudio-notebook extends 'datascience-notebook' with the RStudio development environment. Note that this image derives from 'datascience-notebook' and is therefore not GPU-enabled.

• Flexible CPU, memory, and NVidia GPU allocations, beginning at 2 CPU cores and 4GB RAM dedicated to each student session. Upper limits based on class size, demand, and capacity.

• Course-specific file/dataset distribution (up to 500GB, with larger corpora accepted on a space-available basis) and up to 10GB storage per student. Allocations at the lower end of that range are used for very large classes; please tell us where assignments generate significant per-student output. → Workspaces & Storage

• Tools for assignment distribution / collection / grading, including roster integration with Canvas.

Grade export to Canvas is performed manually. The workflow generates a CSV file which the instructor or TA uploads to the Canvas gradebook. → Exporting the Grades

• Limited support for per-individual minor customizations within our standard software images.

Course-Specific Customization¶

Instructors may supplement standard software images through addition of language modules (e.g. Python or R libraries), new system-level packages (compilers, utilities, etc.), or more extensive modifications to meet the specific needs of their course.

As experts in their curriculum and subject area, the instructor and/or a designated Technical Point of Contact ("TPOC") must take the lead in the installation, configuration, and student use of the new course-specific features. ITS technical staff provide support for customization via 1:1 Consultation (see below).

The instructor and/or TPOC is expected to have basic familiarity with Unix commands such as 'ssh', 'mkdir' and 'chmod' and should be proficient using the intended core platform (e.g. Python or R) in a desktop Mac/PC setting. ITS will provide guidance and basic training regarding the Datahub/DSMLP environment and any system-specific procedures.

Note that 1:1 Consultation availability is limited in the final weeks of each term.

Highly complex use cases (e.g. those incorporating clustered services such as Spark or Postgres, or software not deriving from our Standard Images) are regularly utilized within Datahub/DSMLP, but will require substantially more time and expertise from the instructor/TPOC than ordinary customizations. See "Complex Customizations & Experimental Features" below for guidance.

A course may also pin a specific image so that students are not moved by a quarterly update partway through a project. → Pinning a Workspace

Support & Technical Consultation¶

IT Services provides prioritized support for standard features and functionality via several routes:

Online Documentation offers general information about Datahub and usage of the platform's various features, as well as documentation for known issues, workarounds, and limitations of the environment.

The IT Service Desk can be reached by Phone, Web, or Email for assistance with:

• Outages, Errors, or Malfunctions, e.g.:
• System issues (e.g. the service or components are unavailable)
• User access issues (e.g. an enrolled student can't login)
• Errors or unexpected behavior within standard features (effort is prioritized by the number of courses/students affected and overall impact on instruction)

We ask that students route functionality concerns through their Instructor or TA.

• Requests & Inquiries:
• Requests for system-side configuration or adjustments to course setup, e.g. resource limits, disk quotas, container tags, file ownership/permissions
• Straightforward questions about core features, system capabilities, clarification of documentation, etc.; in-depth support available via 1:1 Consultation.

Response targets:

Effort is prioritized by the number of courses and students affected. Please note in the ticket where an issue affects a whole class or falls on an exam. Urgent problems arising outside business hours may be escalated through the Service Desk.

1:1 Consultation appointments connect instructors, TAs, and Technical Points of Contact with ITS technical experts for real-time guidance or assistance on standard features, course-specific customizations, and complex/experimental features.

Possible topics include:

• In-depth questions regarding Standard/Core Feature configuration or usage
• Discussion not well-suited to ticket-based interaction
• Functional or technical issues outside of the Standard/Core features
• First-time use of a custom container derived from a standard software image
• Reviewing customization options and selection of additional packages/tools
• Resolving functional or technical issues pertaining to course-custom features, e.g. build errors, version conflicts
• Questions/training regarding our customization process, e.g. use of git, Github, tags, Actions, etc., or installation of language and system packages
• Assistance with maintenance following quarterly Standard Software Image updates

Note that 1:1 Consultation availability is reduced in the final weeks of each term, and staffing levels limit courses to a maximum number of hours per term (see "Caveats and Limitations" below.)

Complex Customizations & Experimental Features¶

The compute clusters underpinning Datahub and DSMLP are capable of hosting complex or novel customizations which fall outside ITS' normal bounds of support for a variety of reasons, e.g. installing or integrating them may exercise untested or seldomly-used features of Kubernetes, Docker, or Linux, or student use of the new features may require sophisticated technical expertise or hand-holding.

Examples of available capabilities considered complex or experimental include:

• Instructor/TA containers not derived from a Standard Software Image
• Student-built customized containers (of any derivation)
• Background batch processing and analysis pipelines
• Matlab (Jupyter kernels or Web UI), GNU Octave, or similar complex applications
• Spark clusters
• ArcGIS integration
• Postgres and other persistent services

Incorporation of these or similar features into coursework will require the instructor and/or TPOC to invest significant time ahead of and during instruction, first to become independently familiar with the underlying technologies and then to serve as primary support for their students' activities.

ITS is eager to provide technical guidance for innovation within our services, but without advance agreement cannot become responsible for implementation or usage. We recommend scheduling a 1:1 Consultation with us at least one full quarter in advance of the planned use to discuss feasibility.

Visual Studio Code is not among these. It is supported and in heavy use, via Remote-SSH over a ProxyCommand. → Remote Editor Setup

GPU Access for a Course¶

From Fall 2026, GPU access is managed by a reservation system. The mechanics are documented in GPU Access; what follows is what a course needs.

A course holds a GPU quota — how many of each class it may hold at once — and its students hold Service Unit (SU) budgets which divide that capacity across the roster. Quotas are date-aware: a course's share may be raised for exactly the span of a project deadline and revert on its own afterwards. → Quotas, Cohorts & Availability

Please tell us the deadlines. The quarterly survey asks instructors and TAs about assignment scope, GPU sizes, and — crucially — deadlines. Those responses are what allow us to raise a course's quota for week 9 in advance rather than discover the surge as it occurs.

Managing the course calendar. As workspace managers, instructors and TAs see the whole course calendar and may book on behalf of a student whose project has run into trouble.

Assisting a student who has reached their SU limit. Two routes; the immediate one first:

1. Book on the student's behalf. This takes effect at once and does not draw that student's budget. This is the appropriate action late in the evening before a deadline.
2. Request a budget or limit change by ticket to datahub@ucsd.edu. The change itself is an administrative action, but requesting it is a normal and sanctioned route.

→ Service Units & Budgets · The Six Requests

Waiving a cancellation charge. A student who misses a booked window is assessed a cancellation charge of up to 50% of the booking. Instructors and TAs may waive this charge where warranted. → The Cancellation Penalty

Reports. Manager reports cover reservations by group, peak simultaneous use by class, reserved hours, and effective limits. Note that some of these display cluster-wide data rather than only the course itself. → Managing a Group

Advising students. Two facts account for most of what reaches course staff: launching a GPU session draws on the student's budget whether or not a calendar is opened, and an idle GPU session is reclaimed after roughly 30 minutes. Both are covered on the student pages — Using Datahub in a Course for browser users, and Working from the Command Line for anything requiring a terminal.

Caveats & Limitations¶

We ask all Datahub/DSMLP users to understand and abide by the following:

No Sensitive Data: Datahub is not engineered to protect highly-sensitive data such as clinical records or export-controlled information ("P4" per University of California classification levels) and must not be used for such purposes. Legally- or contractually- protected information ("P3") may be permitted after review; note that depending on the nature of the data, vetting may take 4-6 weeks or longer.

Shared Consulting Resources: at Spring 2026 staffing levels, each course may request up to 6 hours of 1:1 Consulting services.

Scheduled Maintenance: Datahub may be unavailable Tuesdays, 6-8AM for installation of time-sensitive updates or security patches. In practice this work is generally limited to a subset of worker nodes, in which case running jobs are unaffected. Infrequent 'Critical' updates may require downtime outside of this timeframe, in which case we will notify instructors as soon as practical and attempt to minimize impact on coursework and long-running jobs.

Shared Compute Resources: Datahub and DSMLP system resources are shared among all courses assigned to Datahub and DSMLP. Demand for resources, in particular for GPUs, may exceed capacity at peak hours during 10th and Finals Weeks or at assignment deadlines. From Fall 2026, reservations, per-course quotas and Service Unit budgets are the mechanisms by which equitable access is provided in such situations.

Self-Supporting Programs: "Self-supporting" programs (e.g. MAS, MBA, etc.) are welcome to utilize DSMLP/Datahub for coursework or projects, but UC policy requires us to recover associated direct and indirect costs. Please contact us to discuss.

Availability and Reliability: Datahub and DSMLP were designed with student workloads in mind, deliberately trading some of the costly redundancy typical of financial or health settings for additional capacity and capability. As such, they should not be used to host externally-available services or applications except as required for coursework or projects. (This caveat applies primarily to the compute nodes executing student jobs; critical components such as networking, file storage, and backups are maintained to Enterprise IT standards.)

Appropriate Use: The campus-wide IT Acceptable Use Policy applies to use of Datahub and DSMLP, including prohibitions on commercial or political activity, hacking or cyberstalking, and other types of unwelcome behavior.

If you still have questions or need additional assistance, email us at datahub@ucsd.edu, submit a ticket, or call the ITS Service Desk.

Suggest an edit to this page