Autoscaling environments automatically add or remove resources
as demand changes.
About this task
Autoscaling groups have policies that specify when to add or remove resources. For example,
a policy can specify that the group should add an instance when the average CPU utilization of the
existing instances is above 80%. In most cases, you use autoscaling groups with load balancers,
which automatically direct load to the resources in the group.
If there are two or more resources
in an autoscaling group, those resources are stored in a nested blueprint.
Restriction: Autoscaling groups, and the load balancers that manage them, are supported only
for OpenStack clouds. Autoscaling policies are controlled by the Ceilometer infrastructure in
OpenStack clouds. Autoscaling policies are triggered by alarms that are generated by Ceilometer. In
turn, alarms are triggered by samples (data points) that are generated by meters. To learn more
about alarms, samples, and meters, see
Measure cloud resources in the OpenStack documentation.
Results
When you deploy the autoscaling group, OpenStack Ceilometer
monitors the group and adds and removes resources based on the group
policies. The load balancer spreads requests throughout the resources
and directs traffic to resources with available capacity.
The following
blueprint is an example of an autoscaling group with a load balancer.
The blueprint includes a virtual image and component that are independent
of the group. The autoscaling group contains a virtual image and a
component.