When you deploy a blueprint, workload placement uses collected data to recommend where to deploy the blueprint based on available resources. vRealize Automation and vRealize Operations Manager work together to provide placement recommendations for workloads in the deployment of new blueprints.
While vRealize Automation manages organizational policies, such as business groups, reservations, and quotas, it integrates with the capacity analytics of vRealize Operations Manager to place machines. Workload placement is only available for vSphere endpoints.
Workload Placement Terms Used
Several terms are used with workload placement.
- Clusters in vSphere map to compute resources in vRealize Automation.
- Reservations include compute and storage, where the storage can consist of individual datastores or datastore clusters. A reservation can include multiple datastores, datastore clusters, or both.
- Multiple reservations can refer to the same cluster.
- Virtual machines can move to multiple clusters.
- When workload placement is enabled, the provisioning workflow uses the placement policy to recommend where to deploy the blueprint.
Provisioning Blueprints with Workload Placement
When you use workload placement to provision blueprints, the provisioning workflow uses the reservations in vRealize Automation, and the placement optimization from vRealize Operations Manager.
- vRealize Automation provides the governance rules to allow placement destinations.
- vRealize Operations Manager provides placement optimization recommendations according to analytics data.
- vRealize Automation continues the provisioning process according to the placement recommendations from vRealize Operations Manager.
If vRealize Operations Manager cannot provide a recommendation, or the recommendation cannot be used, then vRealize Automation falls back to its default placement logic.
When a developer selects a catalog item and completes the form to request the catalog item, vRealize Automation accounts for the following considerations to provision the virtual machines.
|Policies||The vRealize Automation reservation policy might indicate more than one reservation.|
|Reservations||vRealize Automation evaluates the request, and determines which reservations can satisfy the constraints made in the request.
The reservations are updated to track that resources have been consumed.
If vRealize Operations Manager recommends a cluster or datastore that vRealize Automation considers to be out of capacity or no longer applicable, vRealize Automation logs the exception. vRealize Automation allows provisioning to continue according to its default placement mechanisms.
To identify resources for a virtual machine, vRealize Automation provides a list of candidate reservations. Each candidate in the list can include a cluster and one or more datastores or datastore clusters. vRealize Operations Manager uses the candidate reservations to create the list of destination candidates and locate the best target.
The policy in vRealize Operations Manager sets the level of balance, utilization, and buffer space for the cluster. For a single reservation, which is a cluster or datastore cluster, vRealize Automation validates whether the recommendation is viable placement destination.
- If the destination is viable, vRealize Automation deploys the blueprint according to the recommendation.
- If the destination is not viable, vRealize Automation uses the default placement behavior to place the virtual machines.
Placement considerations must also account for health and utilization problems. While the cloud administrator and virtual infrastructure administrator manage the infrastructure, developers care about the health of their applications. To support developers, the workload placement strategy must also consider health and utilization problems.
|Workload Problem||Placement Solution|
|Developer notices a health problem in the environment.||
vRealize Automation is provisioning blueprints in clusters that are experiencing problems, or that are overutilized because of large workloads. vRealize Automation must integrate with the capacity analytics in vRealize Operations Manager to ensure that blueprints are provisioned in clusters that have sufficient capacity.
|Developer notices a utilization problem.||
The clusters in the environment are underutilized. vRealize Automation must integrate with the capacity analytics that vRealize Operations Manager provides to ensure that blueprints are provisioned in a cluster where the utilization is maximized.
Users Who Provision Blueprints
The following users perform actions to provision blueprints.
Cloud Administrator or Virtual Infrastructure (VI) Administrator
Ensures that the initial placement of virtual machines meets organizational policies, and that they are optimized according to the operational analytics data.
|IaaS Admin role|
Defines the reservations, reservation policies, and placement policy in vRealize Automation.
|Fabric Administrator role, Infrastructure Architect|
Defines the endpoints for vSphere and vRealize Operations Manager, which are necessary for workload placement.
|IaaS Admin role|
As a blueprint architect who works directly with virtual machine component types, assigns the reservation policies to virtual machines when authoring a blueprint. Specifies the reservation policy as a property of the machine component in the blueprint.
|3||Infrastructure Architect, Application Architect, Software Architect, and XaaS Architect||
Creates and publishes the blueprint to provision the virtual machines. Only the Infrastructure Architect works directly with machine components. The other architect roles can reuse infrastructure blueprints in nesting, but they cannot edit the machine component settings.
The blueprint can include a single component, or it can include nested blueprints, XaaS components, multiple virtual machines in a multi-tier application, and so on.
vRealize Automation places the virtual machines according to the configuration of the reservations, and optionally includes the reservation policy at the machine component level for the blueprint. For example, your blueprint might include two machines, with a different policy applied to each machine.
vRealize Automation also optimizes the virtual machines according to the operational analytics data that vRealize Operations Manager provides.
Cloud Administrator or VI Administrator
Selects the policies that govern the initial placement of the virtual machines that vRealize Automation provisions.
The Administrator can:
|IaaS Admin role, Infrastructure Architect|
Builds the custom data center and custom groups in vRealize Operations Manager. Then, the VI Administrator applies the policy used to consolidate and balance workloads to those custom data centers.
|IaaS Admin role, Infrastructure Architect|
Selects the placement policy in vRealize Automation.
Use the workload placement policy to have vRealize Automation determine where to place machines when you deploy new blueprints. The placement policy requires input from vRealize Operations Manager
|Fabric Administrator role|
Requests a blueprint to provision virtual machines.
The blueprint can consist of multiple machines to run a three-tier application.
When the developer deploys the blueprint, vRealize Operations Manager searches for a placement policy that fits the relevant clusters for the request.
For more information about the placement policy, see Placement Policy.
To configure workload placement, see Configuring Workload Placement.