Architecture design practices: Large-scale Distributed System Based on WCF

This article goes to http://hb.qq.com/a/20101119/001701.htm

For applications in large systems, the overall function of an application system with a well-designed architecture must be composed of many functional modules, the data required by each function module corresponds to one or more tables in the database. In architecture design, the more unified and less interaction points between functional modules, the lower the Coupling Degree of the system, and the better the maintainability and scalability of each system module.

In a system, especially a relatively large system, such as an e-commerce system, the functions can be divided, order Management System, commodity management system, user management system, supplier management system... in such an application, the system is split, the function is independent, the coupling is reduced, and maintenance is easy. But how can we complete the communication in each system? If, how does the user system query the order information of a user? Is it necessary to directly reference the DLL of the Order System, or put the functions that need to be called each other in a project called Common for all subsystems to call, or simply query the order database across databases in the user system. These methods are not good. At this time, it is time for Microsoft's powerful WCF to appear. Let's take a look at the architecture diagram below.

Advantages of this design: 1. servers between layers can be flexibly combined. servers in each layer can be horizontally expanded (clusters) and vertically scalable (split by system/domain/function. all middleware services can provide services for this system or provide services for other systems (even third parties such as java), and provide seamless connections between systems, coupling between systems does not depend on applications or databases, but on service interfaces. 3. all operations on the database depend on the middleware and direct operations on the database are not allowed. In this way, the security of the database can be achieved and the Application Server can be reduced, because the actual query is not completed by the data access layer of the application server, it is shared to the middleware server.