UCR Home ITS HPCC Campus status Get help
Research ComputingUniversity of California, Riverside
Home / Knowledge Base / Launching an Ursa Major research workstation
Cloud

Launching an Ursa Major research workstation

This guide shows how to create a research workstation (a Compute Engine VM) in your Ursa Major project. For what a workstation is good for, see Ursa Major research workstations.

Before you start

  • Costs: research workstations are Tier 2. The VM, its disks and any GPUs are recharged to a lab funding source under an MOU. See KB005: Ursa Major service tiers and KB007: Ursa Major recharge workflow.
  • Workstations set up before October 2026: some were set up under the earlier tiers, when workstations were not recharged. If your lab has one, contact research-computing@ucr.edu. Nothing changes without that conversation.
  • Exotic hardware: if you need hardware the HPCC does not have, such as TPUs or Arm processors, ask about Tier 1 exotic hardware instead. It runs through a Research Computing cluster rather than a workstation.
  • GPU work: for GPU work at lower cost to the lab, consider the HPCC.
  • Set a budget so the lab sees spending early. See Creating GCP budgets.
  • Sensitive data: P3, P4 or regulated data needs a review before it is used on a VM. See Security and Data.

Web console

  1. Go to the Google Cloud console and sign in with your UCR account.
  2. Select your project in the project picker at the top of the page.
  3. Open the navigation menu and select Compute Engine, then VM instances.
  4. Click Create instance.
  5. Name, region and zone: give the VM a name and choose a region and zone. Use us-central1 (Iowa), the Ursa Major default, unless your work needs another region.
  6. Machine configuration: choose a machine family and type (CPU count and memory). For GPUs, choose the GPUs machine family and pick a GPU type and count. Not every GPU is offered in every zone.
  7. OS and storage: click Change to pick the operating system (for example Debian, Ubuntu, Rocky Linux or Windows Server) and the boot disk size and type.
  8. Additional disks (optional): under storage, add a new persistent disk if you want data kept separate from the boot disk. Disks are charged while they exist, even when the VM is stopped.
  9. Networking: review firewall and network settings. A public IP address is available by request to research-computing@ucr.edu. Do not open SSH or RDP to the whole internet. See Connecting to an Ursa Major research workstation.
  10. Review the monthly estimate shown on the page, then click Create.

Command line (gcloud)

  1. Install the Google Cloud CLI, or use Cloud Shell in the console.

  2. Sign in and set your project:

    gcloud auth login
gcloud config set project my-lab-project

  3. Create a VM. This example makes a Debian VM with 4 vCPUs and 16 GB of memory:

    gcloud compute instances create my-workstation \
    --zone=us-central1-a \
    --machine-type=e2-standard-4 \
    --image-family=debian-12 \
    --image-project=debian-cloud \
    --boot-disk-size=100GB \
    --boot-disk-type=pd-balanced

    Replace the name, zone, machine type, image and disk size with what you need. For Ubuntu, use for example --image-family=ubuntu-2404-lts-amd64 --image-project=ubuntu-os-cloud.

  4. (Optional) To attach a GPU to an N1 machine type, add these flags. GPU VMs cannot live-migrate, so the maintenance policy must be TERMINATE:

    --accelerator=type=GPU_TYPE,count=GPU_COUNT --maintenance-policy=TERMINATE

    Some GPU types (such as A100, L4 and H100) come with their own machine types (A2, G2, A3) instead. See Google's Create a VM with attached GPUs.

  5. The VM is usually ready within a few minutes. See Connecting to an Ursa Major research workstation.

Google's guide: Create and start a VM instance.

Tips

  • Size to the work. Start small; you can change the machine type later while the VM is stopped.
  • Stop VMs you are not using. A running VM is charged whether or not you are working on it. Disks are charged until deleted.
  • Choose the operating system for your software, and check that it is still supported.
  • Choose storage for your data: boot disk for the system and software, a persistent disk for working data, and a storage bucket for data you want to share or keep after the VM is gone.
  • For long batch runs, the HPCC cluster is usually a better fit. Ask us if you are not sure.
Owner: Research Computing Reviewed: 4 Oct 2026
Related guides
GuideAlphaFold on a Slurm cluster with Google Cloud FilestoreGuideDistributed PyTorch training with Kubeflow TrainerKB002When your Ursa Major project is readyKB005Ursa Major service tiersKB007Ursa Major recharge projects: how setup worksKB021Requesting an Ursa Major (Google Cloud) projectGuideLLM Inference SettingsGuideRunning open-source language models with OllamaGuideLaunching an Ursa Major HPC clusterGuideUrsa Major guidelinesGuideUrsa Major HPC clustersGuideUrsa Major project budgets and spending alertsGuideUrsa Major research servicesGuideUrsa Major research workstationsGuideConnecting to an Ursa Major research workstation