Manage virtual environments with recommendations
Virtualization Manager recommendations focus on the optimization of resource allocation based on performance metrics, historical data, available resources, and storage capacity. Recommendations calculate trends and risks based on enabled strategies, providing plans of action to consider and apply to resolve immediate issues or preemptively prevent issues from occurring.
The Recommendation Engine ensures VM stability and performance through provided actions:
- Balance the virtual environment for optimal performance. You can spread load over hosts, which optimizes the performance for all virtual machines.
- Optimize capacity. Virtualization Manager predicts when you will run out of resources on virtualization entities, so you can plan in advance when you have to assign new resources.
- Predict possible future performance and capacity issues based on past trends.
Recommendations and alerts in VMAN do not require additional configurations to monitor and trigger. When you finish installing and configuring VMAN with your VMware or Hyper-V credentials, data autopopulates into the system providing immediate results for generated events, alerts, and recommendations. You can simply investigate occurring or potential issues, review recommended resolutions, and verify the solutions against reported data within a drill-able page without hunting for stray data across multiple screens and 3rd party managers.
As you take any action on a recommendation or virtual system through the Orion Web Console or 3rd party management (like vCenter), the recommendations entirely recalculate. The Recommendations page and resources always display the latest recalculated list. You can also select to recalculate all recommendations with the option.
Strategies for recommendations
The Recommendations Engine uses a series of strategies to provide specific resolutions for current and potential issues in your VM environment. By using strategies simultaneously, you can achieve optimal balance between performance and efficiency.
VM sizing optimizations
These strategies help you find VMs under and over allocated VM resources. Overallocated VMs have too many resource, best utilized by underallocated VMs. Underallocated VMs do not have enough resources allocated to them to support processes, potentially causing alerts.
Host Performance and Capacity Assurance
These strategies help you find hosts that will run out of CPU or memory resources within a defined time period, and hosts with CPU or memory usage currently over a critical threshold.
Storage Performance and Capacity Assurance
This strategy helps you determine datastores that may reach a storage space or other limit within a defined time period.
Balancing VMs on hosts
This strategy helps you distribute VMs on hosts to achieve balanced usage and utilization percentage.
Migrating VMs is only possible between hosts that are in the same cluster and have shared storage.
Recommendation constraints and policies
Policies define constraints for the Recommendation Engine per selected virtual objects in your virtual environment. These constraint policies include exclusion policies for excluding objects and disallow action policies.
Exclusion policies specify virtual objects to exclude from generated recommendations. You can set up an exclusion policy for specific virtual objects including virtual machines, hosts, clusters, and datastores. When creating an exclusion policy, you select monitored virtual objects in you environment from generating recommendations.
Disallow action policies specify virtual objects to not run additional actions for a recommendation. Recommendations may require actions selected to complete environment and virtual object changes. These changes include moving the VM to a different host or storage and changing resource configurations such as CPU and memory amounts. Options differ between virtual object types.
Active and predictive recommendations
Virtualization Manager provides both active and predictive recommendations
Receiving an alert informs you the VM has an issue, causing problems in your environment and for users. The alert provides specific information for the issue with a details page to research what happened and what to do, such as CPU Contention. You need to take action fast. You could access the alert and review the data and recommended actions to begin determining what to do, or check your Recommendations for a better, faster solution.
Active recommendations trigger for alerts and issues that have already occurred in your environment. These recommendations display with an ACTIVE flag and provide immediate resolutions for alerts based on captured and calculated VM usage data and status. Active recommendations are linked to alerts, providing an intelligent resolution with actionable steps to complete when applied.
Active recommendations provide recommended actions to solve problems that are currently occurring. These recommendations require a minimum of 1 hour of VM monitored data.
Depending on the size of your VM environment, you could receive extremely large amounts of historical and real-time data on performance, status, and allocation usage. You could use complicated formulas to determine what may or will happen based on usage trends, attempting to predict the future of resource allocation vs usage.
As your VM resources are used and managed, Virtualization Manager collects data on VM component usage and status. When reviewed and calculated, the Recommendations Engine uses this data to forecast trends of usage and identify potential problems, including historic information for over and under resource allocation. Recommendations provide intelligent, preemptive actions to prevent issues with VM resources and performance.
Predictive recommendations predict the future state of the virtual environment based on historical trends and data. These recommendations require a minimum of 7 days of VM monitored data. Predictive recommendations may be dependent between other recommendations, as changes completed by one predictive recommendation may affect trends triggering another recommendation.
- Removing load from a host or datastore and adding another load to the same resource
- Changing resources on a VM and then moving the VM to another location
The key to predictive recommendations is compiled data over time. The more VM monitored data accumulated, the engine predicts usage and potential issues.
As you view VMs with alerts and recommendations, drill down further to locate the specific issue and potential solutions. The selected VM encountered an issue, displaying an alert with an associated recommendation. Click to view additional information for the triggered alert and recommendation. The ACTIVE tag indicates this issue has occurred. The recommendation has an immediate resolution you can apply.
Select the recommendation to further review information and make your selections to take action. The Statistics tab allows you to review what actions will take place when applied and a before and after comparison of consumption and usage. All recommendation solutions base entirely on the intelligent decisions of the Recommendations Engine and collected data for the VM.
Recommendations trigger and display with a severity level like alerts: critical and warnings. The severity is assigned based on global and VM specific threshold settings. For example, a percentage of memory, CPU, or capacity reached by 80% triggers a warning severity alert or recommendation.
Other recommendations trigger for predictive actions without using thresholds to set the severity. The importance or amount of environment changes in recommendation steps can affect the severity as warning or critical. For example, a right-sizing recommendation informing you to reduce CPUs on a VM may display as "critical" if it heavily affects the environment or requires a number of changes.
You can take one of the following actions for recommendations:
Perform Now: immediately performs the recommendation steps to resolve the issue.
Example: A VM triggers alerts for disk space utilization for a critical VM for database and running heavy load queries. To respond to the alert and resolve issues, you would perform the recommendation immediately.
Schedule Recommendation: sets a date and time to complete the steps. The Recommendation moves from the Current tab to the Scheduled tab. At the selected time, the steps complete. You may want to schedule the changes if it requires steps that could cause issues during specific times in your environment.
Example: Reports may run at the end of the week, requiring large amounts of VM resources. Moving VMs to new hosts during this time could cause CPU contention or other issues.
Ignore this Recommendation: creates an exclusion policy to not provide recommendations for the affected virtual system according to a set time range. To ignore, click More Actions and select Ignore Recommendation on a single recommendation. You can only ignore one recommendation at a time, not for multi-selected recommendations.
- Create Policies for Recommendations: creates a policy affecting on a single recommendation, such as blocking configuration or move actions.