VMware Aria Operations uses the demand model and allocation model for Workload Optimization. By default, only the demand model is used. You can turn the allocation model on in the active policy.
What is Allocation Model?
The allocation model determines how much compute, memory, and storage resources are allocated to object types. You define the allocation values by modifying the policy which is applied to the objects. The allocation values, also known as overcommit ratios, affect performance and cost.
The allocation model works alongside the demand model. Unlike the demand model which always affects the capacity calculations, the allocation model can be turned on or off in the policy setting. You can control the ratio by which VMware Aria Operations overcommits either the CPU, memory, or disk space. By specifying the allocation values in the policy, you can choose whether you want to overcommit your resources or not. Overcommitting helps you measure utilization of resources in a pay-as-you-go model. When you do not overcommit, the utilization of your cluster will never exceed 100%. If your resource utilization is over the allocation ratio that you set, Capacity Remaining becomes zero.
To modify a policy and configure overcommit ratios, see Policy Allocation Model Element in VMware Aria Operations Configuration Guide.
What is Demand Model?
The demand model is a concept used to predict future resource requirements based on historical data and usage patterns. The demand model is a key component of capacity planning and optimization, allowing administrators to anticipate the growth of workloads and infrastructure resources needed to support those workloads over time.
The demand model takes into account various factors, such as historical resource consumption, trends, seasonality, and potential changes in workloads. By analyzing this information, VMware Aria Operations can provide insights into when additional resources (such as CPU, memory, storage, and network capacity) will be needed to meet the expected growth in demand.
Difference between Allocation Model and Demand Model in Workload Optimization
When clusters in a data center are not optimized, VMware Aria Operations uses the demand model by default to optimize the VMs placed in the cluster. While VMware Aria Operations moves VMs to optimize the cluster, it uses the capacity demand to calculate the best way to optimize the cluster.
When you turn on the Allocation Model in the active policy, VMware Aria Operations optimizes based on the Demand Model, and takes into consideration the overcommit ratios defined by the Allocation Model. As the VMware Aria Operations WLP engine takes into the consideration the overcommit ratios based on the allocation model for CPU, memory and disk space, these resources will not be under stress after optimization due to the over allocation. In any optimization model which is based on allocation the target is always to reduce the stress on all clusters and ensure that after the VM migration the target clusters are not over-allocated.
How to Enable Allocation Model
- Cluster compute resource. Set the overcommit ratios for CPU, memory and disk space. You can also consider powered off VMs by selecting the Activate checkbox.
- Datastore. Set the overcommit ratio for the vSAN datastore disk space.
- Datastore cluster. Set the overcommit ratio for cluster disk space.
After you have set the overcommit ratios, click Save. After setting the overcommit ratios for the Allocation Model in the active policy, go to the Workload Placement page to see the Allocation Model displayed under the Are your clusters meeting your utilization objective? section.