Using Anaconda and Miniconda at SLAC

Table of Contents

Brief Description

This article explains when and how to use Anaconda or Miniconda for scientific computing at SLAC, with a strong recommendation to use centrally managed environments on the SLAC Shared Science Data Facility (S3DF). It also clarifies responsibilities and cybersecurity expectations for any local desktop installations.

Using Conda: Central S3DF vs. Local Desktop

  • Central scientific computing infrastructure (S3DF) is the recommended place to use tools such as Anaconda or Miniconda.
  • Desktops are best suited for personal productivity (for example, reading email) and for connecting to S3DF using tools such as an ssh terminal, NoMachine, or Jupyter.

Recommendation Summary

  • Preferred: Use a group-managed Anaconda installation on S3DF.
  • Acceptable with responsibilities: Install Anaconda or Miniconda on your desktop only if you must, and be prepared to maintain it frequently.

Using Conda on S3DF (Recommended)

  • S3DF provides centrally managed, group installations of Anaconda that reduce your administrative burden and improve security posture.
  • For instructions and options, see Conda on S3DF.
  • Typical access patterns include connecting from your desktop to S3DF via:
    • ssh terminal sessions
    • NoMachine
    • Jupyter services

Local Desktop Installations: Responsibilities and Risks

  • If you choose to install Anaconda or Miniconda on your personal desktop:
  • It is NOT centrally updated. You are responsible for patching and updating it often.
  • It is NOT automatically updated. You must understand and perform routine updates.
  • Expect ongoing system administration time and attention from you personally.
  • These are key reasons why the centrally managed, group Anaconda installation on S3DF is recommended.

Cybersecurity Scanning and Remediation Process

  1. SLAC IT Cyber Security performs routine scans that detect vulnerable Anaconda or Miniconda installations on desktops.
  2. If a vulnerable local installation is found, you will be notified and given the opportunity to update it.
  3. If you choose not to update, your local Anaconda or Miniconda installation will be archived (tarred and gzipped) for disaster recovery purposes.
  4. You may later unzip and untar the archived installation if needed, but the recommended path forward is to use an existing or new group-managed Anaconda/Miniconda installation on S3DF.

Preferred Usage Patterns

  • Use conda within a container on S3DF compute nodes and batch infrastructure whenever possible.
  • Connect from your desktop to S3DF (for example, via ssh, NoMachine, or Jupyter), and run your scientific workloads centrally.

Getting Help

  • Discuss your Anaconda/Miniconda needs with the S3DF Scientific Computing Support team using the Conda on S3DF page.
  • SLAC users may also ask questions in the #comp-sdf Slack channel.