Rackspace Auto Scale Glossary

This article defines the following common Auto Scale terms:


A monitoring daemon that resides on the server you are monitoring. The agent
gathers metrics based on agent checks and pushes them to Rackspace Monitoring.

Agent token

An authentication token used to identify the agent when it communicates with
Rackspace Monitoring.


A mechanism that contains a set of rules that determine when a notification is


The act or process of confirming the identity of a user or the truth of a claim.
The authentication service confirms that the user who claims to be making an incoming
request is actually the one making the request. The service does this by
validating a set of claims that the user makes. These claims are initially in
the form of a set of credentials. After initial confirmation based on
credentials, the authentication service issues a token to the user. When making
subsequent requests, the user can provide the token as evidence that the service
has already authenticated the user's identity.


A definition that explicitly specifies how you want to monitor an entity.


Software that collects data from the monitoring zone. The collector maps
directly to an individual computer or a virtual machine.


The act of Auto Scale adding or removing enough servers to satisfy the needed

Convergence Delta

The change in the number of servers that the system makes when a scaling policy
is executed. For example, if the convergence delta is 2, then the system adds two
servers. If it is -10, the system removes ten servers.


See Group cooldown and Policy cooldown.


A resource that you want to monitor. Some examples are a server, a website, or a


A resource configuration for a server. Each flavor is a unique combination of
disk, memory, vCPUs, and network bandwidth.

Group cooldown

The length of time that must pass before a scaling group can be scaled up or
down again. The cooldown prevents events from triggering a new policy execution
before server builds initiated by a previous policy execution complete.

Health monitor

A configurable feature of each load balancer. A health monitor is used to
determine whether a back-end node is usable for processing a request. The load
balancing service currently supports active health monitoring.


A collection of files for a specific operating system (OS) that you use to
create or rebuild a server. Rackspace provides prebuilt images. You can also
create custom images from servers that you have launched. Use custom images
for data backups or as golden images for additional servers.

Launch configuration

A configuration that contains the necessary details for adding and removing
servers from a scaling group in the Rackspace Auto Scale API. The
launchConfiguration object specifies whether you are creating a server or a
load balancer and the necessary details about the configuration.

Load balancer

A logical device that belongs to a cloud account. You can use a load balancer to
distribute workloads between multiple back-end systems or services, based on the
configuration criteria.


A back-end device that provides a service on a specified IP and port.


An informational message that is sent to one or more addresses when an alarm is

Policy cooldown

The length of time that must pass before a policy can be executed again. The
purpose is to allow for a fast scale-up and a slow-scale down of servers.


The process of adding or reducing capacity in response to changes in workload.

Scaling group

A set of identical servers, and optionally a load balancer, that can scale up and
down in response to load, as defined by the scaling policy and bound by the
scaling-group configuration.

Scaling policy

A policy that specifies how much capacity (that is, cloud servers) to add or
reduce. A schedule-based scaling policy also specifies when this should happen.
An event-based scaling policy relies on alerts that trigger webhooks.

Session persistence

A feature of the load balancing service that attempts to force subsequent
connections to a service to be redirected to the same node as long as the node
is online.


A computer that provides explicit services to the client software running on its
system. A server is a virtual machine (VM) instance in the Cloud Servers
environment. To create a server, you must specify a name, flavor reference, and
image reference.

Server image

See Image.

Virtual IP

An Internet Protocol (IP) address that you configure on the load balancer.
Clients use the virtual IP to connect to a load balanced service.
Based on its configuration, the load balancer distributes incoming connections
to back-end nodes.


A URL that can activate a specific (scale-up or scale-down) policy for a scaling
group without requiring authentication.