The following considerations should be applied while using this pattern:

• Planning: Decision to throttle requests affects the design of system. Throttling strategy and resources that need to be throttled should be decided on before implementation because it is difficult to add throttling after the system has been implemented.
• Response: A clear response indicating that user request has been throttled should be sent back to the client. This would help the client retry an operation after some time.
• Auto-scale versus Throttle: If there is a temporary spike in application load, then only throttling may be used instead of scaling the application.