Load Testing Tools
The Azure Local Load Testing Framework includes five standalone load testing tools. Each tool targets a specific performance domain — storage, network, database, or system stress — and can be run independently from a management workstation with no CI/CD pipeline required.
Choosing a Tool
flowchart TD
A[What do you need to test?] --> B{Storage I/O}
A --> C{Network}
A --> D{Database}
A --> E{CPU / Memory / Burn-in}
B -->|Windows HCI host\nFleet-scale| F[VMFleet]
B -->|Linux VMs\nFine-grained control| G[fio]
C --> H[iPerf3]
D --> I[HammerDB]
E --> J[stress-ng]Tool Summary
The Status column is the same for every tool today. The per-tool detail (implemented, unit-tested, proven on hardware) is in the implementation status.
| Tool | What It Tests | Target OS | Run From | Status |
|---|---|---|---|---|
| VMFleet | Storage IOPS, throughput, latency via DiskSpd VM fleet | Windows (HCI host) | PowerShell | Implemented, unit-tested, not proven on hardware |
| fio | Storage I/O with fine-grained control (block size, queue depth, I/O engine) | Linux VMs | PowerShell + Ansible | Implemented, unit-tested, not proven on hardware |
| iPerf3 | Network bandwidth, jitter, packet loss between nodes | Linux / Windows | PowerShell | Implemented, unit-tested, not proven on hardware |
| HammerDB | SQL Server OLTP (TPC-C) and OLAP (TPC-H) benchmarking | Windows VMs | PowerShell | Implemented, unit-tested, not proven on hardware |
| stress-ng | CPU, memory, cache, and I/O stress for burn-in and capacity planning | Linux VMs | PowerShell + Ansible | Implemented, unit-tested, not proven on hardware |
Running a Test (Standalone)
Every tool follows the same three-step pattern. No pipeline or CI/CD setup is needed — just a PowerShell terminal on your management workstation.
Step 1 — Install the tool on target nodes
.\tools\<tool>\scripts\Install-<Tool>.ps1 `
-ClusterName "hci01.corp.infiniteimprobability.com" `
-Nodes @("hci01-node1", "hci01-node2")Step 2 — Run a test with a workload profile
.\tools\<tool>\scripts\Start-<Tool>Test.ps1 `
-ClusterName "hci01.corp.infiniteimprobability.com" `
-Nodes @("hci01-node1", "hci01-node2") `
-Profile "general"Step 3 — Collect results
.\tools\<tool>\scripts\Collect-<Tool>Results.ps1 `
-ClusterName "hci01.corp.infiniteimprobability.com" `
-Nodes @("hci01-node1", "hci01-node2") `
-RunId "<run-id-from-previous-step>"Each tool page below has the exact commands with real parameters.
Common Lifecycle
All tools share the same eight-phase lifecycle. When running standalone, you execute each phase manually. The CI/CD pipelines automate this same sequence if you want to schedule or trigger runs automatically.
| Phase | What Happens | Manual Command |
|---|---|---|
| Pre-Check | Validate cluster connectivity and prerequisites | Initialize-Environment.ps1 |
| Install | Deploy tool binaries to target nodes | Install-<Tool>.ps1 |
| Deploy | Configure test environment (VMs, volumes, etc.) | Deploy-<Tool>.ps1 (VMFleet only) |
| Test | Execute workload profile | Start-<Tool>Test.ps1 |
| Monitor | Collect PerfMon / Azure Monitor metrics (parallel) | Watch-<Tool>Monitor.ps1 |
| Collect | Parse and aggregate raw results | Collect-<Tool>Results.ps1 |
| Report | Generate PDF / DOCX / XLSX | New-TestReport (ReportGenerator module) |
| Cleanup | Remove test artefacts (optional) | Remove-<Tool>.ps1 |
Automation (Optional)
If you want to automate test runs on a schedule or as part of a CI/CD workflow, see CI/CD Pipelines. The pipelines call the same scripts listed above — they are not a separate system.
Diagrams
Architecture and workflow diagrams are maintained as draw.io files in docs/assets/diagrams/ and exported as PNG images to docs/assets/images/ for documentation use.
| Diagram | Source | Description |
|---|---|---|
| Solution Architecture | solution-architecture.drawio | End-to-end framework overview |
| VMFleet Workflow | vmfleet-workflow.drawio | VMFleet pipeline phases |
| Config Flow | config-flow.drawio | YAML → JSON configuration pipeline |
| CI/CD Pipeline | ci-cd-pipeline.drawio | GitHub Actions / Azure DevOps flow |
| Network Topology | network-topology.drawio | Cluster network paths for testing |
NOTE
Exporting Diagrams To export updated PNGs from draw.io source files, the MkDocs drawio plugin handles this automatically during site builds. You can also export manually from draw.io desktop or VS Code extension.