Every Ember application is represented by a class that extends [`Ember.Application`][1].
This class is used to declare and configure the many objects that make up your app.

As your application boots,
it creates an [`Ember.ApplicationInstance`][2] that is used to manage its stateful aspects.
This instance acts as the "owner" of objects instantiated for your app.

Essentially, the `Application` *defines your application*
while the `ApplicationInstance` *manages its state*.

[1]: http://emberjs.com/api/classes/Ember.Application.html
[2]: http://emberjs.com/api/classes/Ember.ApplicationInstance.html

This separation of concerns not only clarifies the architecture of your app,
it can also improve its efficiency.
This is particularly true when your app needs to be booted repeatedly during testing
and / or server-rendering (e.g. via [FastBoot](https://github.com/tildeio/ember-cli-fastboot)).
The configuration of a single `Application` can be done once
and shared among multiple stateful `ApplicationInstance` instances.
These instances can be discarded once they're no longer needed
(e.g. when a test has run or FastBoot request has finished).