Table of Contents |
---|
Panel | ||
---|---|---|
| ||
This page describes basic scaling group concepts and actions. |
Introduction to autoscaling
To configure an automatic response to changing demands for resources, you can scale out VMs or scale them in, which is also called horizontal autoscaling. To scale out, the platform clones the base VM and deploys the clones. To scale in, the platform will delete clone VMs and undeploy the base VM. Scaling operations are subject to all standard platform constraints, such as privileges and allocation limits.
...
Limitations:
Autoscaling does not clone captured VMs, so to use scaling groups with a captured VM, create an instance and recreate the VM. Create instances to save Save VM disks to templatescreate an instance template
VApp specs do not support scaling groups. See What do virtual appliance specs save and create
Scaling groups have aggregate alarms that are associated with the base VM. This means that you can push custom metrics for clone VMs but you cannot create alarms for cloned VMs that are part of a scaling group.
State of base VM: A scaling group with a deployed base VM would be destroyed if the base VM were deleted directly on the hypervisor. In contrast, a scaling group with an undeployed base VM is not vulnerable to interference at the hypervisor level
...
Go to Virtual datacenters → Virtual appliances
On the VM icon, from the options menu, select Define scaling group
Enter the scaling parameters
For the Default cooldown, enter the period of time to wait from the start of one scaling operation before allowing another scaling operation
For the Minimum running virtual machines that Abiquo must maintain in the scaling group, the value must be greater than or equal to zero, where zero means that the base machine is not deployed
The option to Keep virtual machines in the same layer can maintain VM anti-affinity layers when autoscaling
Administrators with the privilege to Manage workflow for scaling groups can Disable workflow or enable it as required
Optionally, select Create in maintenance mode to delay the start of autoscaling, and the automatic deployment of VMs to meet the minimum size
Select the option to Create autoscaling action to create basic operations to scale in and scale out, with triggers based on metrics and alarm conditions.
Create scaling rules
For Scale out rules, enter the number of VMs to Add. This isthe number of times to clone the base VM and deploy each clone for each scaling step
For Scale in rules, enter the number of VMs to Remove. Abiquo will delete clone machines and undeploy the base machine
If there is no time range, then this is a default scaling rule. A time range must be unique and cannot overlap with other rules with the same scaling direction.
Click Save
...
When you define a scaling group, select Create autoscaling action and Save the scaling group
In the dialog, select a Metric to control an autoscaling action
To configure more options, including the thresholds for scaling in and scaling out, click Show more
To add this action, click Add
Add more actions as required
...
Create a VM and a scaling group for the VM. See Define a Manage scaling groupgroups
If you create an automatic scaling action, then the VM metrics will trigger autoscaling when they cross the thresholds set for the actions
...
Create an action plan with a scaling action for the VM with the scaling group. See Create an action plan to automate VM actionsManage action plans
Create triggers to run the action plan. See Create a trigger for an action planManage action plans
When scaling, the platform will search for a scaling rule that is valid for the specific time range, or for a default rule. It will create or delete/undeploy the number of VMs in the rule, then wait for the cooldown period before accepting another scaling request.
...
Go to Virtual datacenters → Virtual appliances → select VM
At the bottom of the VM icon, click the cog maintenance symbol at the bottom of the VM icon
OR if the scaling group is open, click the spanner maintenance symbol in the top right corner
To leave maintenance mode
...