Bemærk
Adgang til denne side kræver godkendelse. Du kan prøve at logge på eller ændre mapper.
Adgang til denne side kræver godkendelse. Du kan prøve at ændre mapper.
Compare Chaos Studio Workspaces and Experiments (classic) to choose an Azure Chaos Studio resource model for each resilience test. Workspaces is the current model: it discovers resources within a scope, runs Scenarios, and generates Scenario reports. Experiments (classic) is the legacy model being replaced by Workspaces. With Experiments (classic), you enable each target individually and combine faults to build custom tests. Workspaces is in public preview; Experiments (classic) is generally available.
Important
Chaos Studio Workspaces and Scenarios are in public preview. Microsoft provides this preview "as is" and "as available," and it isn't covered by service-level agreements or limited warranty. Microsoft provides customer support for the preview on a best-effort basis. This preview isn't meant for production use. For more information, see the following articles:
How the two models differ
Both models inject real faults against your Azure resources, but they differ in how you get from "I want to test this failure" to a running test.
| Aspect | Chaos Studio Workspaces (preview) | Experiments (classic) |
|---|---|---|
| Servicing state | Active development. New Chaos Studio features ship in the Workspaces model. | Legacy model being replaced by Workspaces. There's no further feature development, and only critical fixes, such as security updates, are considered for backport. |
| Onboarding | Set a scope (subscription, resource group, or service group). The Workspace discovers supported resources automatically. | Enable a target and capabilities on each resource before it can be used in an experiment. |
| Test definition | Start from a Scenario template that already composes the Actions and sequencing for an outage pattern, or customize one in the Scenario designer. | Assemble faults, steps, and branches manually, and select target resources for each fault. |
| Finding what to test | The Scenario library shows which Scenarios apply to the resources discovered in your scope. | Choose faults from the fault library for Experiments (classic) and check resource requirements. |
| Identity and permissions | Configure managed identity at the Workspace level: system-assigned, user-assigned, or both, with role assignments shared by its Scenarios. The portal offers to fix missing role assignments. | Use a managed identity and role assignments for each experiment. |
| Permission validation | The Workspace validates the permissions required for a run before it starts. | Missing permissions typically surface as failures at run time. |
| Regions | A Workspace is a logical resource that can act on resources in any Azure region from any of the supported Workspace regions. | Experiments deploy to specific regions, and targets must be in a resource-targeting region. |
| Reporting | Each run produces a downloadable Scenario report with run details, an Action summary, a timeline, and an execution flow diagram. | Experiment history shows execution details and error information per run. |
| Supported tests | Use the curated Scenario catalog or customize a Scenario in the designer or as a Microsoft.Chaos/workspaces/scenarios resource with the Actions available for custom Scenarios. |
Use the classic fault catalog, including agent-based faults, AKS Chaos Mesh faults, and dynamic targeting. |
Main advantages of Chaos Studio Workspaces
Faster onboarding. A Workspace removes the largest source of setup friction in Experiments (classic): per-resource target and capability enablement. You set a scope once, and discovery finds the supported resources in it. If you add or remove resources later, the Workspace picks up the changes automatically.
Built-in recommendations. The Scenario library is populated from the resources discovered in your scope, so you start from outage patterns that apply to your environment instead of assembling faults manually. Scenarios such as Compute Zone Down compose related disruptions for you.
Use any region. Workspaces are logical resources: the Workspace region doesn't need to match your target resources, and one Workspace can act on resources in any Azure region. In Experiments (classic), targets and capabilities must be co-located with your resources, and experiments can only be created in certain regions.
Simpler, safer permissions. The Workspace's managed identity is a single blast-radius control shared by every Scenario, with a two-layer authorization model: the person triggering the run needs permission on the Workspace, and the Workspace identity needs roles on the target resources. When validation finds missing permissions, the portal offers to fix the role assignments for you. In Experiments (classic), you manage an identity and role assignments for each experiment.
Validation before disruption. A Workspace validates that the identity has the permissions a run requires before the run starts, so permission gaps surface before Actions run.
Evidence you can share. Every run generates a structured Scenario report with run details, Action statuses and durations, affected resources, a timeline, and an execution flow. You can download the report for compliance reviews, post-incident retrospectives, or resilience maturity assessments.
When to choose Chaos Studio Workspaces
Choose Workspaces when:
- You're new to Chaos Studio and want the fastest path from nothing to a meaningful resilience test.
- The outage pattern you need is covered by the Scenario catalog or is close enough to customize in the designer.
- You need to test resources across multiple regions or subscriptions from one place.
- You need structured, shareable evidence of testing for compliance or operational-resilience frameworks.
- Multiple teams or applications need separate testing boundaries. You can create a Workspace for each application, environment, team, or compliance boundary.
When to choose Experiments (classic)
Experiments (classic) is being replaced by Workspaces. During the transition, choose Experiments (classic) only when:
- You need a fault that the scenario catalog doesn't yet cover, such as specific agent-based faults or AKS Chaos Mesh faults for in-cluster fault injection.
- You rely on classic-only capabilities such as dynamic targeting or scheduled experiment runs.
- You require a generally available resource model. Chaos Studio Workspaces is in public preview and isn't recommended for production workloads.
Microsoft no longer adds features to Experiments (classic) and considers only critical fixes, such as security updates, for backport to this legacy model. Plan to move each test to Workspaces when it meets your capability and general availability requirements.
Move from Experiments (classic)
Follow Move from Experiments (classic) to Chaos Studio Workspaces to map your experiments to Scenarios, run equivalent tests in a Workspace, and clean up classic resources. Don't assume that a Scenario tests the same resources and faults as your experiment. Check each replacement Scenario in a preproduction environment first.
Use both models during the transition
You can run both models side by side as you move your tests. Use Chaos Studio Workspaces when the Scenario catalog covers the outage you want to test and preview use meets your requirements. Keep an experiment in Experiments (classic) if it needs a generally available resource model or a classic-only fault or capability.