Because the backend may provide various interfaces, and developers may access these interfaces in various ways when writing node-side code. If we don't handle the interface access mode and usage in a unified architecture, it will bring the following problems:

1. Every developer uses his own code style to write interface access code, which makes the project directory and coding style confusing and relatively difficult to maintain.

2. Each developer writes his own simulation data pattern. After the development is completed, he needs to manually modify the code to delete the mock.

3. In order to switch interfaces in different environments (daily, pre-release and online), each developer may maintain some configuration files.

4. The data interface call pattern is not easy to be reused by every service model.

5. The description convention of data interface is scattered in every corner of the code, which may be inconsistent with the interface document agreed by the back-end personnel.

6. After the whole project is developed independently, the cost of interface debugging or test regression is still high, which needs to involve every interface provider and user.