Park: stop or start cloud resources

Overview

Parkable entities

Currently, Turbonomic supports parking actions for the following entities:

Some entities that display in the Parking page may be unparkable when you view them because the entities apply a parking policy that disables parking actions. The Unparkable column indicates if such a policy currently applies. These entities cannot be parked on demand or automatically.

For example, you may have virtual machines in your production environment that must be running all the time. You can prevent the enforcement of parking actions by creating a parking policy that disables parking actions.

How parking works

Cloud workloads that become idle for a period of time (for example, after business hours) and those that undergo regular maintenance are ideal for parking. When parking actions are enforced, Turbonomic reverses any conflicting actions on these entities. For example, when a VM has been started via Turbonomic and a user stops the VM via the cloud provider, Turbonomic will start the VM after a subsequent discovery cycle.

When a scaling action for a VM or Database Server is in progress, Turbonomic postpones a stop or start action until the scaling action completes.

Enforcing parking actions

There are several ways to enforce parking actions.

• Execute actions on demand.

  Use the Parking page to execute parking actions on individual entities. For details, see Executing Parking Actions.

• Attach a parking schedule to one or several entities.

  Use the Parking page to attach a parking schedule to one or several entities. You must create parking schedules as a prerequisite. For details, see Managing Parking Schedules and Scheduling Parking Actions.

  Note:

  When you execute a parking action on demand and the affected entity has an existing schedule, a system prompt notifies you to pause the schedule before executing the action. This prevents potential conflicts between on-demand and scheduled parking actions.

• Create parking policies.

  Use parking policies to enforce actions dynamically and at scale. For example, you can create a parking policy at the cloud provider level so that all existing and newly discovered parkable entities from a specific cloud provider are parked uniformly.

  Use the Policy Management page to create parking policies. For details, see Parking Policies.

Parking policies take precedence over the other enforcement methods. For example, if a parking policy dictates that a VM must be running everyday from 9:00 AM to 6:00 PM, and a Turbonomic user manually stopped the VM from the Parking page at 3:00 PM, the VM will be started to comply with the parking policy.

Working with the Parking page

After discovering parkable entities, Turbonomic adds them to the Parking page and shows their state (stopped or running) at the time of discovery. Use the Parking page to execute parking actions or attach parking schedules to these entities.

Users with the Site Administrator, Administrator, Automator, or Parker role can manage the parking actions and settings in the Parking page. Users with other roles have read-only access to the page. For details about roles, see Managing User Accounts.

General tasks:

• For a long list of entities, you can type a search keyword or add filters to narrow the results.

  For filters, valid values display in a dropdown list. These filters support regular expressions and partial keyword matching to help you narrow down the values further.

• To set an entity as your scope, click the entity name.

• To download the data in the Parking page, click the download button.

Be aware of the following as you manage parking actions.

• To see and download a list of parking (and non-parking) actions that have been executed successfully, add the All Actions chart to your dashboard and use the tabular view.

• Actual savings associated with parking actions are reflected in the Savings and Cumulative Savings charts, but not in the Parking page.

  For an entity that is currently running, the Cost column in the Parking page shows the hourly on-demand rate for that entity, for your reference.

Details for AWS Auto Scaling Groups and Azure Virtual Machine Scale Sets

The table in the Parking page shows individual AWS Auto Scaling Groups and Azure Virtual Machine Scale Sets, but not the member instances. To view member instances, click the information icon in the table.

A new page opens and includes the following tabs:

Executing parking actions on demand

Parking actions can only be executed from the Parking page. The Pending Actions, Potential Savings, and Top Accounts charts in the Home Page only show the non-parking actions that you can execute.

To execute a parking action, select one or several entities with the same state, and then click Stop or Start. In the following example, the selected entities are currently stopped and will be started.

In the All Actions chart, the tabular view shows the following information for the parking actions that you executed:

• Execution: Manually Accepted

• Risk Type: Savings (stop actions) or Efficiency Improvement (start actions)

Scheduling parking actions

Attach a parking schedule to one or several entities to ensure that parking actions occur automatically at the appropriate times. You must create parking schedules as a prerequisite. For instructions on creating parking schedules, see Managing Parking Schedules.

To attach a schedule, select one or several entities without a schedule and then select Attach from the dropdown menu.

In the Parking Schedule page, choose from a list of previously configured schedules.

If no schedule is suitable or available, create one by clicking Create schedule. Be aware that a schedule created this way will immediately apply to the entities in scope. If this is not your preference, create a new schedule by navigating to Settings > Schedule and then clicking New Schedule > Parking Schedule.

After attaching a schedule, the name of the schedule appears in the Schedule column.

Click the schedule name to view the settings for that schedule. To edit the settings, navigate to Settings > Schedule and then open the schedule for editing.

In the All Actions chart, the tabular view shows the following information for scheduled parking actions that have been executed:

• Execution: Automatically Accepted

• Risk Type: Savings (stop actions) or Efficiency Improvement (start actions)

Pausing a parking schedule

You can pause schedule enforcement for a given period of time. For example, you can pause during non-typical work hours or when an unanticipated maintenance window requires automation outside Turbonomic to start and stop workloads. If the entity was running, it keeps running. If it was stopped, it stays stopped.

To pause a parking schedule, select one or several entities with a schedule, and then select Pause Schedule from the dropdown menu.

In the Pause Schedule page, specify the duration of the pause.

In the Parking page, an icon appears before the name of the schedule to indicate that it has been paused.

The parking schedule will resume automatically when the pause period is over. To manually resume the schedule, select the affected entities and then select Resume Schedule from the dropdown menu.

Detaching a parking schedule

You can detach parking schedules anytime. Select one or several entities with schedules, and then select Detach Schedule from the dropdown menu.

Notice for updated Turbonomic instances

If you updated your Turbonomic instance from version 8.7.0 (or older) and noticed that the parking feature is missing in the user interface, perform the following steps in your updated instance:

1. Open charts_v1alpha1_xl_cr.yaml for editing.

   vi /opt/turbonomic/kubernetes/operator/deploy/crds/charts_v1alpha1_xl_cr.yaml

2. In the suspend section, change the value for enabled to true.

     # Suspend service selector
     suspend:
         enabled: true

3. Apply the change.

   kubectl apply -f /opt/turbonomic/kubernetes/operator/deploy/crds/charts_v1alpha1_xl_cr.yaml

4. Restart the UI and API pods.

   kubectl delete pod <ui-podname>

   kubectl delete pod <api-podname>