There is no difference in performance beween staging and production. In fact, in a VIP swap all that happens is that the load balancer redirects traffic from the instances currently in the production slot to the instances currently in the staging slot -
and then swaps the slot names.
The idea with staging is that you use it for smoke testing your application prior to doing a VIP swap. It is not intended for routine testing of a hosted service. The best practice is to use a separate hosted service for testing - which helps avoid the accidental
promotion of a test service to production. The Microsoft Patterns and Practices Team
book on Moving Applications to the Cloud has the following premise for its application life cycle chapter:
-- Adatum have a well-defined set of processes for deploying applications to their on-premises servers. They use separate servers for testing, staging, and production. When the development team releases a new version of the application, its progression through
the testing, staging, and production servers is tightly controlled. Very rarely, though, small fixes, such as updating the text on an ASPX page, are applied directly to the production servers.
The Patterns and Practices books on
developing applications for and
moving applications to the cloud are both available on the MSDN website.