Atmosphere Platform Decommissioned: The History of Atmosphere and Alternative Options for Users

Nov. 23, 2022

While CyVerse's Atmosphere platform has been decommissioned, CyVerse offers powerful and versatile alternatives to meet your cloud and compute needs.


After 11 years of continuous operations, CyVerse's Atmosphere, a cloud platform for scientific analysis using virtual machines, has been decommissioned, but with this change, exciting new powerful and versatile CyVerse platforms have taken its place. CyVerse discontinued Atmosphere for general analysis use cases on September 1, 2020 and completely decommissioned it for all users in 2022. This change came about primarily due to the changing landscape for performing scientific analysis in the cloud.

Atmosphere was conceived and developed in 2010 by two CyVerse staff members, Seungjin Kim and Edwin Skidmore, with a public launch for CyVerse users in 2011. In 2014, the NSF-funded Jetstream program adopted Atmosphere to provide a unified view of cloud resources across Jetstream's multiple clouds and enabled researchers the ability to easily manage virtual machines in the cloud.

Atmosphere developer and Director of Infratructure at CyVerse, Edwin Skidmore, said, "Although Atmosphere's end of life marks the end of an era for doing research in the cloud, we are excited to see Atmosphere's design philosophy live on in other open-source research projects as well as the platform that the CyVerse Cloud Native team is currently working on." He notes that CyVerse and its collaborators are building an exciting new service, called CACAO, from the lessons learned through helping researchers do compute and storage in the cloud, including:

  • The ability to use multiple types of clouds
  • The ability to create Jupyterhubs and other complex software stacks at scale
  • The ability to use the scientific frameworks that are ideal for researchers

CACAO, which is available through Jetstream2, enables greater capabilities and accessibility to clouds than Atmosphere. CACAO, which stands for Cloud Automation and Continuous Analysis Orchestration, was created to address the evolving needs of researchers to perform their analyses across different cloud providers, both commercial (i.e., AWS, GCP, Azure) and institutional (i.e., university clouds or NSF ACCESS CI). Using CACAO, users can do their scientific analyses regardless of where their data resides or funding agency constraints (i.e., NIH STRIDES, NSF Cloudbank). They can also monitor and control their cloud costs by deploying CACAO to their teams and organizations.

Users will still be able to utilize the CyVerse Discovery Environment for most needs, including the ability to manage their data and analyses in one place. With widespread adoption of software containers (i.e., DockerSingularity) for distributing applications (such as BioContainers), the CyVerse Discovery Environment is a web-based orchestration platform and workbench for managing data, performing analyses, and visualizing results (see Run Analyses with the Discovery Environment for more information). CyVerse also provides extensive training and documentation for customizing and scaling container-based analysis. Most users will be able to migrate their virtual machine-based analyses to containers.

Additionally, users can run interactive analysis through CyVerse's Visual and Interactive Computing Environment (VICE) in the Discovery Environment. Users who don't currently have access to VICE will need to request access ( VICE allows users to interact with their data and perform analyses in their favorite programming language in one place with an interactive format. VICE also provides a cloud shell for users who may need a simple way to execute Linux command-line tools. Additionally, there is a Linux desktop-based VICE app. Users who still require virtual machines can utilize CACAO.

We welcome your collaboration opportunities around the use of CACAO and cloud automation for your organization and projects. Questions about Atmosphere or CACAO can be directed to

Create Account

An Open Science Workspace for Collaborative Data-driven Discovery