One big bank deepening system second: design strategy

One, blending technology (MASHUP)

Mashups are a blend, an interactive Web application that leverages content retrieved from an external data source to create new innovative services. Mashups are essentially a variety of web applications that use public or private databases, which are combined to form an integrated application by invoking the content provider's Web APIs. Mashups typically use the API interface of the source application, or some RSS output (including atom) as a content source; Unlike portal portals, mashups can consolidate pages and content without needing a portal container.

Mashup Web sites are characterized by the use of content and functionality obtained from data sources outside the organization's boundaries, and by fusing and assembling them on the client or server side. The mashup mechanism reduces the coupling of the system and provides a way to share the business from the underlying data to the upper pages.

Second, queuing theory (queue theory)

By using queuing theory, the performance prediction, performance analysis and performance evaluation of work item pool and instance pool are carried out, and the specific parameter configuration can be calculated to improve the efficiency of the system through quantitative evaluation of the system.

Through the statistical analysis and process modeling analysis, find the key factors affecting the operation of the system, so as to guide the design, optimize the system.

Third, pipelining technology (Pipe line)

The pipeline technology is used in the system modeling, and the aim is to provide the theoretical basis for the evaluation and analysis of system performance, throughput and efficiency.

Iv. Modular Design (Component Based)

The component design greatly strengthens the system's robust and maintainable, each component is relatively independent, the component calls based on the advertisement interface Convention, the system local change will not affect the whole, thus the enterprise may replace the old component to satisfy the work need according to the development need.

V. Services Registration and Discovery (Service registery/discovery)

In an SOA system, service producers create services, encode or configure them to perform some kind of business process, and publish directory information for services registered with the service. Reusing in an SOA means that many service consumers share the same service instance. Because when a shared service is stopped or ported to another place, the application system that uses it terminates, so it is important to ensure that the service consumer does not terminate when the service provider's implementation and location changes.

Service registration is based on the UDDI standard. UDDI Services Registration is open, xml-based, platform-Independent, allows the business to be displayed on the Internet, and defines parameters for interacting with other business interactions, or a group within the enterprise develops its services to other groups. The UDDI registration mechanism is a directory based architecture whose registration includes technical models and business models.

Vi. interface Programming (programming to Interface)

Interface programming is primarily a design idea that separates functions from implementations to reduce coupling and maintain the relative independence between components. By providing different subclass implementations, the system has better scalability, while increasing the stability and robustness of the code, reducing the coupling and so on.

Vii. one-way dependence (N-tier architecture Model)

Multi-tier design usually increases the complexity of some systems, but it helps to improve the scalability of the system and the manageability/maintainability of the system. The aim is to make the system easier to understand, the different parts are relatively independent, and can be easily replaced and improved. Attention should be paid to reducing the coupling of layers and layers, where we use one-way dependencies to reduce coupling between layers, while using IOC containers to manage these dependencies.

Viii. design pattern (GOF Classic mode, Java EE core mode, EJB design mode)

In object-oriented programming, software programmers pay more attention to the reusability and maintainability of previous code. Design patterns make it easier and easier for people to reuse successful design and architecture. The presentation of proven technologies into design patterns also makes it easier for new system developers to understand their design ideas. Can effectively improve the development efficiency and quality, improve the stability and flexibility of the system.

Ix. reversal control of the IOC (Dependency Injection Dependency injection)

The use of IOC, by the container control program in the dependency relationship, rather than in the traditional practice, by the program code active direct control. Control is transferred from the application code to the external container. The caller does not have to understand the creation of the callee or even create the callee itself, thereby enabling the caller and the callee to decouple. It also facilitates the container to provide other crosscutting services, such as transaction control.

X. AOP (Aspect Orient programming)

There are some questions (aspects/concerns), such as Access/Shutdown database connection, transaction management, security control, etc., generally exist in multiple modules or components, AOP provides crosscutting ideas, across different components, the problem provides a centralized unified solution, simplifying development, reducing coupling, is a useful complement to OOP.

Xi. UML-based Ooa/ood

UML is an industrial standard and is widely used in system analysis and design. Ooa/ood, based on object-oriented thinking, conforms to the custom of OOP developers and provides a good foundation for OOP. Ooa/ood combined with UML can improve the efficiency and effectiveness of communication and improve the quality of analysis and design.

12. Serialization (serialization)

The mutual conversion mechanism between object and byte stream. Through the serialization operation can complete the Memory object network transmission, the binary storage, the relational database management and so on function.

Note http://blog.csdn.net/leverage_1229