Skip to main content

Limits & supported configurations

reference · Reviewed 2026-07-30 · FEAT-0035 · FEAT-0036 GA

What is supported, what is not, and the numbers to plan around. Support varies by cloud provider and operating system, so check here before committing to a configuration.

Operating systems​

OSSupported
Windows Server 2019✅
Linux✅

Cloud providers​

ProviderWorkstationsSuspensionReboot
AWS✅✅✅
Microsoft Azure✅✅✅
Oracle Cloud (OCI)Limited — Linux only, contact Rescale❌❌
Google CloudContact Rescale——

Region availability differs per provider. If you need a specific region, confirm with your Rescale representative before planning around it.

Display protocols​

ProtocolSupportNotes
In-browser✅No setup; click Connect
Native client✅Requires the client and TCP port 8443
QUIC / UDP✅Native client only — never in-browser
VNCLimitedContact Rescale; not covered by standard support
Microsoft RDP❌Not supported

Timings​

OperationExpected
Resume from suspensionUnder 3 minutes
First launch — Linux~15 minutes
First launch — Windows~25 minutes
Suspend (shutdown)~5 minutes
Reboot — AWS2–3 minutes
Reboot — Azure15–20 minutes

Windows launches can exceed 25 minutes with some hardware and software combinations. See why start-up is slow.

Latency​

SetupExpected
In-browserUnder 100 ms
Native client + nearby regionUnder 50 ms

Suspension eligibility​

RequirementDetail
NodesSingle-node only — multi-node cannot be suspended
Storage100 GiB per core or less
Volume typePersistent storage only, not ephemeral
Hardware change on suspendNot supported
Resume regionMust be the same region it launched in

Ineligible coretypes show "Hardware not eligible for Suspension" in the hardware settings. See suspend and resume.

Session duration​

Workstations are limited to 30 days, counting active and suspended time together.

This is a soft limit, Rescale won't force-stop your workstation at 30 days, but past that point it isn't guaranteed to work. A workstation is a snapshot of its software and worker code at launch, and after a month that snapshot is stale enough that health checks can start failing.

If a long-running workstation is terminated this way, files are usually recoverable, but running past 30 days is at your own risk. If you need a setup indefinitely, save it as a template and launch fresh periodically.

Pricing components​

ComponentApplies toBasis
ComputeAll nodesVaries by coretype
Platform licenseAll nodesPer core-hour
Workstation licenseHead node onlyPer core-hour

While suspended, compute and workstation license stop; the platform license continues on every node. See what a workstation is.

Priority models​

PrioritySupport
On-Demand Priority (ODP)✅ Default
On-Demand Reserve (ODR)✅ Linux
On-Demand Economy (ODE)❌ Not supported
Why interruptible pricing isn't offered for workstations

On interruptible (spot) capacity, the provider can reclaim your machine at any moment. For a batch job that's recoverable. It reruns. For a workstation there is no way to save your data when it's reclaimed mid-session.

That's why ODE is disabled for workstations. If it's enabled by exception for your organization, understand that data loss is unrecoverable and start-up time guarantees no longer apply.

Not supported​

Status
Adding software after launch❌ — unless Administrator Privileges is enabled
Changing hardware on suspend❌ — planned
Hibernation (state to memory)❌ — use suspension, which saves to disk
Desktop pools❌
Microsoft RDP❌
Multi-node suspension❌
Reboot/reprovision on Workstations Basic❌
Audit-log export API❌ — UI only

Workstations Basic​

A limited tier without suspension or collaboration features. It's transitional, organizations on it move to full Workstations at renewal. If you can't find suspension or sharing, check with your admin whether you're on Basic.