Skip to content

Migrate from AWS Research and Engineering Studio Support ends September 2027

AWS support for Research and Engineering Studio (RES) ends on 30 September 2027. This page maps RES features to RosettaHub and sets out the technical differences.

Example migration plan from AWS RES, October 2026 to September 2027: RES keeps running with AWS support until 30 September 2027 while projects move to RosettaHub one at a time, each with its own budget; take stock, map and connect; bring images and data; run the first project in parallel; move the rest; retire RES.

What AWS announced

AWS has ended active development of Research and Engineering Studio on AWS. See AWS's announcement.

Date What happens
29 September 2026 Version 2026.09 is the final release
30 September 2027 AWS technical support and security updates end. RES stays open source (Apache 2.0) and keeps running in your account, but maintaining it becomes your responsibility

Since 29 September 2026, AWS no longer develops RES or supports new users; RES stays open source, and support for its final release ends on 30 September 2027 (see AWS's RES end-of-support notice). RosettaHub keeps your research environments on AWS, in your own accounts, with an account and a budget per project.

From RES to RosettaHub, feature by feature

Everything below is in production today. The desktops, clusters and storage run in your own AWS accounts.

What you use in RES In RosettaHub
Linux and Windows virtual desktops Virtual desktops in the browser, started by the researcher
Software stacks Custom images: any researcher, not only an administrator, can save an environment as an image and share it with a project, a group or other institutions
Project file systems Shared file systems (Amazon EFS, Amazon FSx) per project, and a file browser to upload and download. See File Storage
HPC alongside desktops HPC clusters launched per project, with GPU instances where the project allows them
Notebooks and IDEs Jupyter, RStudio and VS Code in one click. See Jupyter Lab
Single sign-on Your institution's identity provider. See Identity and Access
SSH access SSH to project machines without researchers handling keys
Projects and project roles Project membership and roles managed by the principal investigator. See Projects
Desktop sharing Share a running session with a colleague. See Sharing
Desktop schedules and allowed instance types Idle desktops stopped automatically or on a schedule; instance types allowed per project

Technical differences

RES behaviour is from AWS's RES documentation.

RES RosettaHub
Platform Deployed and maintained in your AWS account; no AWS support or security updates after 30 September 2027 Run by RosettaHub (SaaS or dedicated); environments run in your AWS accounts
Project isolation Projects inside one RES environment One AWS account per project, with its own service quotas
Cost attribution res:Project cost allocation tags, which can take up to two days to become available By account; FOCUS 1.3 and CUR 2.0 data enriched per user and project, in your own account
Budget enforcement AWS Budgets on project tags; when a budget is exceeded, new virtual desktops are blocked, and running desktops continue Cost estimated from resource events in near real time, AI included; at the limit, new activity is refused and running services are stopped
Caps Budget total per project Hourly cost cap, and caps on machines, volume size and storage
Software Software stacks prepared by administrators Custom images any user can save and share
Identity and collaboration Users from the environment's directory Institutional single sign-on; a person can belong to several organizations, so partners join projects without new accounts
Access paths The RES portal Browser desktops and labs, federated AWS console and CLI, API, CLI, SDKs and MCP server
Budget transfers Budgets set by administrators in AWS Budgets Principal investigators move budget between their projects; transfers are logged and unused budget carries over
Administration Administrators and project owners within one environment Institution, department, research group and project, each administered at its own level. See Organizations
Policies and compliance Instance types and software stacks chosen per project Regions, services, instance types and AI models per role, user and project; continuous compliance implemented with Cloud Custodian
Landing zone Not part of RES A multi-account structure with FinOps, audit and log-archive accounts, kept in step with your organization. See The Landing Zone

Migrating, one project at a time

Start well before September 2027, and keep RES running until every project has moved.

  1. Take stock. List your RES projects, users, software images, file systems and budgets.
  2. Map and connect. Map each project to a RosettaHub project with its own AWS account and budget. Onboard your organization and connect your single sign-on.
  3. Bring images and data. Rebuild your software images as custom images, and attach or copy the shared file systems.
  4. Run one project in parallel. Move a first project and let its researchers work in both for a few weeks.
  5. Move the rest. Move the remaining projects one at a time, each with its own budget from day one.
  6. Retire RES. Switch RES off before AWS support ends on 30 September 2027.

Common questions

Is RosettaHub a drop-in replacement for RES?

No product is. RosettaHub covers what researchers relied on RES for, and you move projects across one at a time, with RES running alongside until the last one has moved.

What happens if we keep running RES after 30 September 2027?

RES keeps running in your AWS account, but AWS no longer provides technical support or security updates. Patching, upgrades and fixes become the responsibility of your team or a partner.

Do we stay on AWS?

Yes. Your desktops, clusters, file systems and data run in your own AWS accounts, in the Regions you choose, at your own AWS rates. RosettaHub governs those accounts from its platform; your data stays where it is.

Can production research services be protected from budget stops?

Yes. Enforcement is chosen per account: experiments and sandboxes stop at their budget, while production services can be monitored and alerted on without ever being stopped.