We have two Jupyter environments available through CCR. Both would be suitable for hosting teaching, depending on course needs. One, Open OnDemand, allows users to run Jupyter sessions on our high-performance computing cluster. The second, VIDIA, allows users to run Jupyter sessions on a dedicated CCR server. Arrangements must be made ahead of time to host a course on our resources so that it receives proper support. We need to register all students, ensure the resource will be available at the course meeting time, and provision any requested software packages. For compute cluster and/or OnDemand access, faculty register a class in our ColdFront portal. VIDIA access for coursework is handled by email@example.com.
At this time, we charge the relevant departments a base level for staff support time involved in course support. If more substantial development is needed, it would be priced separately and subject to staff availability.
1. Open OnDemand for Jupyter:
Open OnDemand is a web-based portal to cluster computing access. Jupyter may be run on the CCR cluster with a Python2 or Python3 kernel, and users can request the number of cores, the walltime, and the cluster partition to run on. Users request the session using a web-based interface. They are subject to walltime limitations (72 hours) and queueing for requested resources, as all cluster users are. Once the session starts, it looks just like any other Jupyter session, but it's running on the cluster. OnDemand can be accessed by any currently registered CCR user (all UB faculty, staff, and students are eligible to register). More information can be found here: OnDemand
2. VIDIA for Jupyter:
VIDIA is a web-based portal. Users can use a web browser to access it and run (and even develop) various computational tools. These tools execute in dedicated virtual machines, directly on the VIDIA server. Jupyter users on VIDIA can run Python2, Python3, Octave, or R kernels. For example, we support UB Computational Physics I and II on VIDIA--using Jupyter, among other tools. The platform easily supports 25 or so simultaneous tool sessions. Prospective users can register and have a look around. VIDIA is found here: vidia.ccr.buffalo.edu