Introduction to Design Automation
Design Automation in OneIQ (accessible through the Design menu) is a premium feature that helps you create infrastructure designs based on real-world performance data. Instead of relying on manual calculations or guesswork, it uses observed workload metrics to guide accurate sizing and modeling.
With Design Automation, you can:
Define hardware requirements that match the actual performance of your workloads.
Apply constraints such as CPU type, memory usage (allocated vs. consumed), and power state (powered on or off).
Select specific workloads—at the cluster, host, or virtual machine level—to include in the design.
Model different solution scenarios to compare outcomes and avoid common sizing mistakes.
This feature is intended for users who need to design efficient, cost-effective solutions with confidence that the results reflect the workloads in their environment.
Selecting Workloads
The first step in the Design Automation process is selecting the workloads to include in the model.
How to Select Workloads
Open the Entities menu.
Choose one of the following: Clusters, Hosts, or Virtual Machines (VMs).
Review your selection to ensure no entities are hidden. If needed, click Reset selection to clear the list.
Click Transform to move to the next step.
Should I size based on clusters, virtual machines, or hosts?
Clusters: Use this option to size all virtual machines (VMs) within a selected cluster.
Virtual Machines: Choose this option if you need to model a subset of VMs, such as excluding workloads that are powered off or scheduled for decommission.
Hosts: Recommended when working with Hyper-V clusters, since VM-level data in these environments may be less reliable.
Should I size based on clusters, virtual machines, or hosts?
Clusters: Use this option to size all VMs within a selected cluster. Be aware that cluster-level resources beyond the VMs are also included. These may represent overheads relevant to the current cluster but not required in the new design.
Virtual Machines: Use this option to focus only on VM workloads. It is also useful when modeling a subset of VMs, such as excluding workloads that are powered off or scheduled for decommission. This is the recommended sizing method for VMware, Nutanix, and cloud solutions.
Hosts: Use this option when working with Hyper-V clusters, where VM-level data may be less reliable.
🔹 VMware Environments: Only Clusters or VMs can be added to Design Automation. If you select VMware Hosts, you'll notice the Transform button will be disabled and greyed out.
🔹 Mixed Hyper-V & VMware Hosts: If working with both, create two workloads:
• Hyper-V: Select either the cluster or hosts.
• VMware: Select either the cluster or individual VMs.
2. Design Scenario Settings and Defaults
Once you’ve selected your workload entities and clicked Transform, you’ll be prompted to configure scenario settings.
Key Scenario Settings
Scenario Name: Choose a meaningful name for easy reference. Scenarios are user-specific and not visible to other team members.
Currency: Used for estimated list pricing. These values do not reflect vendor-specific discounts.
Platform Selection: Choose from supported reference hardware configurations based on your premium license type or use a generic option that can be manually mapped to the hardware of your choice.
Origin: Defines whether the data is based on the current date or a historical snapshot (if using the Time Machine feature).
Historical Time Range: Sets the look-back period for data used in sizing.
Base and Peak Rollups & Base-to-Peak Ratio: Solutions are sized based on observed performance, using the midpoint (50%) between the base rollup (average utilization) and the peak rollup (maximum utilization).
vCPU-to-Core Ratio: Defines the maximum number of virtual CPUs per physical core. The default is 6:1 for newer processors, with lower ratios recommended for high-performance workloads such as databases.
Exclude Powered-Off VMs: Omits powered-off VMs when calculating vCPU-to-Core ratios, which reduces estimated hardware requirements.
Use Cloud & VM Insights: Applies right-sizing recommendations from the Insights section (e.g., adjusting over-allocated memory and vCPUs).
Use Memory Allocated for Sizing VMs: Base sizing on consumed memory rather than configured memory, which may lower hardware requirements.
Year-Over-Year Growth Factors: Adds projected growth rates to account for future scalability.
Click Create to generate the design scenario.
Adding Multiple Workloads
You can design for multiple workload types simultaneously, which is useful for complex environments. If adding a new workload to an existing scenario, select Existing Scenario instead.
How to Add Multiple Workloads
Select the first workload type (e.g., VMware Virtual Machines) and click Transform.
Complete the scenario details and click Create.
To add a second workload:
Navigate to Design → Workload and click New Workload.
Return to the Entities view. Click Reset selection before choosing a second workload type (e.g., Clusters or Hosts).
Click Transform again.
When prompted in the Scenario Wizard, select Existing Scenario to merge workloads.
Example Use Case
A VMware cluster and a Hyper-V cluster can be modeled in the same design scenario by creating separate workload entries for each.
Customization and Constraints
Once the scenario has been created and a solution has been generated, you can apply customizations and constraints.
Customization Options
Customization settings are found under the Platforms and Settings tabs.
Platforms Tab
• Displays the platforms OneIQ is designing for.
• Adjust or refine platform options to match customer requirements.
Customization Options (depending on license)
Platform: Displays the selected hardware platform for sizing.
Country: Defines the procurement country, which can affect pricing and parts availability.
Acceptable Lead Time: Filters parts based on availability.
Flexible: Broadest range of parts, may include longer lead times.
No restrictions: Prioritizes in-stock parts.
Processor Vendor & Generation: Specify Intel or AMD processors and the preferred generation.
Storage Type: Choose between Hybrid, All Flash, or All-NVMe.
Model Selection: Allows specifying a particular hardware model if required.
Processor Selection: Choose a processor if you need to meet specific performance requirements.
Software Stack: Select a supported virtualization or cloud software stack (for example, VMware options). This choice may affect hardware requirements.
Applying Constraints
While constraints help tailor solutions, excessive restrictions can limit cost-optimized options. Common constraints include:
Limiting platforms to a preferred brand
Restricting processors to a specific vendor (e.g., Intel only).
Defining a minimum hardware model based on customer preference.
Filtering storage types to match existing infrastructure.
Avoiding Common Selection Errors
Issue: Unexpected Entities in Design Automation
If selecting Clusters or Hosts in the Entities View, you may include more entities than intended due to hidden selections.
Solution:
Click Reset selection before clicking Transform.
Re-select the intended entities (Clusters, Virtual Machines, or Hosts).
Example: Hidden Entity Selection
Before Reset: Two clusters appear selected, but 401 entities are included.
After Reset: Only the intended clusters remain selected.
Can I Combine Multiple VMs Across Platforms?
Yes, but consider the following:
Virtual machines from AWS, VMware, and Nutanix AHV can be consolidated.
Hyper-V statistics may be unreliable, so include Hyper-V hosts as a separate workload.
Understanding Sizing Calculations and Constraints
OneIQ performs sizing calculations based on multiple constraints to ensure the designed solution meets practical requirements. These constraints determine the number and size of nodes or servers recommended.
By default, all sizing uses an 80% utilization threshold and includes N+1 resilience. This means additional hosts may be recommended compared to raw workload numbers. If results seem higher than expected, review these thresholds in the scenario settings.
vCPU Ratio
For many years, a 4:1 vCPU-to-core ratio was standard. With modern processors, 6:1 is often a good balance between cost efficiency and performance and is the default in OneIQ. The setting can be adjusted in the scenario configuration.
The right ratio depends on workload type:
CPU-intensive workloads (e.g., large databases, analytics) may require a lower ratio, sometimes as low as 2:1.
Lightweight workloads (e.g., test or development) can tolerate higher ratios.
Always aim to balance CPU and memory so neither resource is oversubscribed, while leaving headroom for CPU spikes, which fluctuate more than memory usage.
Compute Unit
Different CPUs deliver varying performance per core, making direct comparisons unreliable. To address this, OneIQ uses a standardized measure called a Compute Unit (CU), similar in concept to a processor’s SPECint rating.
A CU is an internal calculation. Each CPU is assigned a performance score and normalized against the Intel Xeon Processor E5-2609 v2 (10M Cache, 2.50 GHz) as the baseline.
1 CU represents the performance of a fully utilized single core of this reference processor.
CUs do not map directly to physical cores; values are adjusted relative to the baseline so CPUs from different generations can be compared consistently.
In sizing, CUs reflect the actual work performed by virtual machines, rather than a generic vCPU ratio. This provides a more accurate sizing method. The solution is then validated to ensure the chosen vCPU-to-core ratio is not exceeded.
Keep in mind: setting the vCPU ratio too low may artificially constrain the design; too high may risk contention. A balance should be maintained.
Memory
Memory sizing is more straightforward than CPU because a gigabyte is a standard unit of measure.
By default, OneIQ uses consumed memory for calculations.
You can switch to allocated memory in the scenario settings by selecting the checkbox when creating the scenario.
All memory sizing applies the 80% utilization threshold and N+1 resilience, ensuring clusters remain balanced and workloads protected.
Storage
Storage is considered in all environments but is only directly sized in hyperconverged platforms such as VMware vSAN and Nutanix, where storage is integrated into the compute cluster.
For platforms using external SANs or in generic compute sizing, storage is assumed to be provided separately and is not included in automated calculations.
Conclusion
OneIQ’s Design Automation streamlines the process of modeling infrastructure for various workloads by providing intelligent, optimized configurations based on real-time data.
By following these steps—selecting workloads, configuring scenarios, adding multiple workloads, and applying constraints—you can efficiently create right-sized solutions that minimize costs while meeting business objectives.