The NMFS Google Workstations Platform

Jonathan Peake

Terminology and Definitions

Terminology used throughout this tutorial are defined below.

Term Definition
Workstation Pre-configured virtual machines listed under “My Workstations” available on NOAA’s Google Cloud.
Configuration The default settings of the workstation including: type (Base, RStudio, Python, posit), and storage/processing size (small, medium, large).
Session An active portion of the workstation, shares storage and power. It is possible to partition a workstation into multiple sessions with different IDEs and core usage.
Data Bucket Cloud-based object storage drive that is optimized for code and external to the workstation.

Workstation Features

  • Persistent Storage: Ensures work saved in your home directory is retained across sessions.

  • Web-Based Access: Enables users to work from any approved device without local installation.

  • Pre-Configured Environments: Standardizes tool availability (R, Python, etc.) for streamlined development.

  • IAM-Based Security Policies: Enforces access controls to safeguard resources.

  • Consistent & Replicable Environments: Provides uniform configurations to enhance reproducibility.

The NMFS Google Workstations “Freemium” model

  • A comprehensive range of standard workstation configurations are provided at no direct cost to the user or their FMC.

  • This service is centrally funded by the OCIO Fisheries Cloud Program and is designed for common data analysis, modeling, and general development workloads.

  • This model ensures all users have access to a valuable baseline service without immediate financial constraint.

This tiered structure allows us to offer a valuable baseline service to all users while providing a scalable path for those with more demanding computational needs.

Workstation ConfigurationsWorkstation Configurations

Workstation Type Recommended Use Case

Base Workstations

(base-*)

 Advanced users who require full control over their computing environment to install and configure custom tools.

Code OSS Workstations

(oss-*)

 General-purpose coding in Python, R, or other languages; similar to VS Code.

Python Workstations

(nmfs-jupyter-*)

 Fisheries scientists and analysts using Python-based tools such as Jupyter for machine learning and geospatial analysis.

R Workstations

(nmfs-rstudio-*)

 Fisheries scientists and analysts using custom R environments for stock assessments, modeling, and statistical analysis.

Posit Workbench Workstations

(posit-*)

 Users working primarily with R and Python in a professional RStudio Pro, VS Code, or Positron environment. Ideal for data scientists and statisticians.

Workstation Sizes

Name vCPUs RAM (GB) Disk (GB) Typical Workload
Small 2 8 10 Single-threaded processing, low memory requirements.
Medium 8 32 50 Mid-tier option for heavier data processing and parallel tasks.
Large 16 64 100 Multi-threaded, memory-intensive workloads and complex modeling.

Which workstation is right for me?

Considerations when choosing a workstation configuration and size:

  • Are you comfortable on the command line?

  • Do you use one coding language over another?

  • Do you need specialized libraries or packages?

  • Does your code run in parallel?

  • Would your code run on your laptop as is?

Customizing workstation environments

All NMFS Google Workstations currently run on a Linux base. You can install any packages or libraries that you need for your workflow. You should be comfortable with running basic commands from a Linux-style command line (e.g., bash terminal) if you plan to customize your workstation outside of “typical use”

Install R packages using pre-compiled binaries, like the Posit Package Manager

Run the following lines in your R console, then restart your R session

repo_line <- 'options(repos = c(CRAN = "https://packagemanager.posit.co/cran/__linux__/jammy/latest"))'
writeLines(repo_line, "~/.Rprofile")
# Restart R session to lock in Rprofile changes
# Session -> Restart R or Ctrl+Shift+F10

Customizing workstation environments

Install R and Python packages at the user level to ensure they persist across sessions

Remember, system directories are ephemeral, but user directories are persistent

install.packages("package_name", lib="~/Rlibs")
pip install --user package_name

High-Performance and Custom Workstations

You can submit requests for workstations with computing capabilities beyond what we have available through the accelerator program (GPU, high-cost compute). These are handled as independent GCP projects and are subject to FMC cost recovery model.

  1. Submit a specific request to: Joshua Lee (joshua.lee@noaa.gov) and Ed Rodgers (ed.rodgers@noaa.gov). 

  2. This request must detail your requirements (machine type, GPU need, estimated usage).

  3. If approved, your FMC’s administrative/financial point of contact will be required to provide the corresponding Organization, Project, and Task number to enable direct billing.