Is it right for you?
A good fit
- A Linux machine you fully control (sudo), for hours or for months
- A GPU machine with part of, or all of, an NVIDIA A100
- Web apps, databases, Jupyter or RStudio servers with a public address
- Classes and workshops that need many identical VMs
- Software that does not suit a batch cluster
Not a fit
- Any protected data (HIPAA, FERPA, PII, CUI and similar)
- Large multi-node MPI jobs (use the HPCC or an ACCESS batch system)
- The only copy of irreplaceable data
- A permanent production service with no allocation behind it
What it is
Jetstream2 is a research cloud funded by the NSF and run by Indiana University. You create your own Linux virtual machines (instances) in a web browser and get full administrator rights on them. Sizes run from one core to 64 cores, and GPU sizes come with part of, or all of, an NVIDIA A100. Each VM can have a web desktop, a web shell and SSH access.
Jetstream2 is allocated through NSF ACCESS. Time is counted in service units (SUs): a VM uses SUs for every hour it runs, at a rate set by its size, and uses none while shelved.
How it fits with other UCR compute
Use the HPCC for batch jobs on campus. Use NRP Nautilus for containers and GPU jobs on Kubernetes. Use Jetstream2 when you want a whole machine of your own: a server that stays up, a GPU you can log in to and work on interactively, or a set of identical VMs for a class.
Data policy
Jetstream2 is for open research data. Do not put protected data on it (see the data protection levels). You administer your VMs, so keeping them patched and locked down is your job.
How to get started
Read the researcher guide to Jetstream2. It covers getting an ACCESS project (see also KB009). Then follow Your first Jetstream2 VM, and GPU VMs on Jetstream2 if you need a GPU. Research Computing can look over an ACCESS request with you. Contact us.