Table of contents
topic
On Enterprise Application Integration
topic introduction
Enterprise Application Integration (EAI) is an approach, methodology, standard and technology for sharing and exchanging information and collaboration between various heterogeneous systems, applications and data sources inside and outside the organization. The applications connected by enterprise application integration include various e-commerce systems, enterprise resource planning systems, customer relationship management systems, supply chain management systems, office automation systems, database systems, data warehouses, etc.
Please discuss the topic of "Enterprise Application Integration" from the following aspects:
- Briefly describe the enterprise application integration project you participated in (project background, initiator, purpose, project characteristics, etc.).
- Briefly describe the four levels (methods) of enterprise application integration.
- Describe in detail which level of integration is used in the project you participated in, how it was implemented, and what the effect is.
Thesis points
Key points for question 1 :
- The outline of the software system: the background of the system, the initiator, the purpose, the development cycle, the delivered products, etc.
- The role of "I" and the main work.
Key points for question 2 :
- Four levels of enterprise application integration.
Key points for question 3 :
- Which level of integration to use.
- The specific implementation process and application effect.
Preparation of theoretical materials
Indicates integration :
Presentation integration is also called interface integration, which is the primitive and shallowest level of integration, but it is also commonly used. This method regards the user interface as a common integration point, and concentrates the original fragmented system interface in a new interface. Indicates that integration is a black-box integration, without knowing the internal structure of programs and databases. Commonly used integration techniques mainly include screen capture and input simulation techniques. Indicates that integration usually has the following situations:
- Configure a PC-based user interface on an existing terminal-based application system.
- Provide users with an application system that looks unified but is composed of multiple systems.
- When integration is only possible on display interfaces. It means that the implementation of integration is very simple, and it is not thorough. It is just a layer of "external decoration", and the extra integration interface may also become the performance bottleneck of the system.
Data Integration :
In order to complete control integration and business process integration, the integration problem of data and database must be solved first. Before the integration, the data must first be identified and cataloged, and the metadata model must be determined to ensure that the data is distributed and shared in the database system. Therefore, data integration is white-box integration. There are many different middleware tools that can be used for data integration. For example, batch file transfer, that is, file transfer between the original system and the newly developed application system in a specific or predetermined way; ODBC standard interface for accessing different types of database systems; databases that provide connections to distributed databases Access middleware technologies and more. Typically, data integration is used when:
- Comprehensive analysis and decision-making of data generated by multiple information sources is required.
- To handle some common repository that multiple applications need to access.
- When it is necessary to obtain data from a data source to update another data source, especially when the data formats between them are different. Relatively speaking, data integration is more flexible than representation integration. However, data integration faces difficulties when the business logic changes frequently.
Control integration :
Control integration, also known as function integration or application integration, integrates application systems on the business logic layer. The integration points that control the integration are stored in the program code, and the integration point may be accessed simply by using the public API, and of course, additional code may be required to realize it. Control integration is black box integration.
When realizing control integration, it can be realized by means of remote procedure call or remote method call, message-oriented middleware, distributed object technology and transaction processing monitor. Control integration is more flexible than presentation integration and data integration. In the environment where integration and data integration are applicable, both are applicable to control integration. However, since the control integration is carried out at the business logic layer, its complexity is higher. Moreover, the business logic part of many systems does not provide APIs, which makes integration more difficult.
Business Process Integration :
Business process integration, also known as process integration, goes beyond data and systems and consists of a series of workflows based on standards and unified data formats. When integrating business processes, enterprises must define, authorize, and manage the exchange of various business information in order to improve operations, reduce costs, and increase responsiveness. Business process integration not only provides interconnection between underlying application support systems, but also realizes end-to-end business process management between applications within the enterprise and between the enterprise and other partners. It includes application integration, B2B integration, automated business process management, manual process management, enterprise portal, and management and monitoring of all application systems and processes, etc.
model essay
Summary
In February 2020, a commercial bank where I work started the construction of a retail CRM project . This project mainly realizes functions such as customer management, customer analysis, marketing management, performance management, data ETL processing, and multi-channel data services . In this project , I act as an architect , responsible for the overall architecture design of the project. This article takes the retail CRM project construction as an example, and mainly discusses the specific application of enterprise integration architecture design technology in this system. The operational CRM and analytical CRM subsystems realize interface integration through the retail CRM portal ; the data service system realizes application integration with each requesting system through the bank's existing ESB ; realizes data integration by using the data bus and data warehouse , and establishes retail CRM data in the warehouse The bazaar provides data support for customer analysis, performance management, and data services. Through the above integrated architecture technology, multiple subsystems in the project were integrated to form an organic whole. Finally, the project went online smoothly. So far, the system has been running stably and has been well received by users.
text
The branches of a commercial bank I work in are located in provincial capitals and key cities across the country, with nearly 60 million retail customers and about 5,000 account managers. The situation of customer management and maintenance is uneven, and the strategies, methods, and tools used by account managers to manage and maintain customers are not unified, which increases operating costs and reduces customer experience. The effectiveness and targeting of marketing activities are not strong, and the marketing costs are often high but the effect is not obvious. In response to these problems and to adapt to the highly competitive market environment, the bank formulated a retail CRM business strategy and launched it in February 2020 The retail CRM project construction is planned , which mainly realizes customer management, customer analysis, marketing management, performance management, data ETL processing, multi-channel data service and other functions. The project aims to establish a unified retail customer management system and retail data analysis environment for the whole bank. , give full play to the business value of long-term accumulated data, scientifically and effectively guide customer acquisition, marketing, service, retention and other life cycle activities, improve customer management and service levels, enhance customer experience, enhance customer value, and thus increase retail business profits return. (Background of the project)
In this project, I acted as an architect and was responsible for the overall architecture design of the project. The project involves a wide range. I adopt the "divide and conquer" strategy. According to the application type, user role, and data processing characteristics, the project is divided into multiple subsystems, and the key technologies of enterprise integration are studied. According to the actual situation of the project, the appropriate integration is selected. technology and architecture design. (Project Enterprise Integration Technology Overview)
Enterprise integration refers to the use of platforms and technologies such as application servers and mid-prices to connect various application systems of the enterprise to achieve interaction and collaboration between heterogeneous systems, as well as data exchange and sharing. There are mainly four types of enterprise integration modes and technologies: 1. Interface integration , which integrates the interfaces of various application systems into one interface, and the common technology is enterprise portal. 2. Process integration , so that various application systems can be connected to support a complete business process. The commonly used technologies are workflow and enterprise portal. 3. Application integration , which provides close to real-time integration of data and programs in two or more application systems. Commonly used technologies are remote procedure call, message middleware, service bus (ESB), Web service, etc. 4. Data integration , which solves the problems of data synchronization between information systems (including the synchronization of the main data system to the copy system, and the synchronization of the source system to the data warehouse) and timeliness (including real-time, batch) issues. Commonly used technologies are adapters and message intermediates software, data bus, data warehouse. Next, I will use enterprise portal, service bus, data bus, data warehouse and other technologies or platforms to realize interface application and data integration to discuss the implementation process in detail. (The four-level theory of enterprise application integration)
1. The operational CRM and analytical CRM subsystems are integrated through the retail CRM portal . The four functions of customer management, marketing management, customer analysis, and performance management belong to the management and analysis requirements for internal users of the bank, and it is suitable to use the B/S structure to build a retail CRM system. In order to reduce coupling, the retail CRM system is divided into two subsystems of operational CRM (realizing customer analysis and performance management functions). However, the convenience for users cannot be reduced. It is necessary to integrate the interfaces of the two subsystems together, which requires the use of interface integration technology, so the retail CRM portal was developed for interface integration. As a unified workbench for retail users, the retail CMR portal provides a unified login interface and access entrance for the two subsystems, unified user authentication and authorization, unified interface style, unified function menu and to-do items, and uses proxy technology to realize the portal system to the two subsystems. The routing and jumping of each subsystem page improves the user experience and makes the division of subsystems transparent to users. The portal system is developed using the J2EE open source framework (Spring, MyBatis). The technical architecture is divided into four layers: page presentation, access control, business logic, and data access. (The operational CRM and analytical CRM subsystems are integrated through the retail CRM portal)
2. The data service system realizes application integration with various requesting systems through the bank's existing ESB . The multi-channel data service function can not only query customer transaction data through the web interface of the retail CRM system for account managers, but also query transaction data for customers through online banking and mobile banking systems. After analysis, the query function can be defined as a standard service, which supports multiplexing and provides interface access for multiple requesting systems. The technology of each system is heterogeneous. It is suitable to use the SOA architecture style to build a data service system, and use the application integration technology to utilize the bank's existing A service bus (ESB) provides query services to multiple systems. ESB includes functions such as service management, protocol conversion, format conversion, service routing, and message processing. Register and publish the query service of the data service system in the ESB, each requesting system initiates a server request to the ESB, ESB access requests, converts the communication protocol and message format through the adapter technology, and submits it to the data service system through service routing query request. (The data service system implements application integration with each requesting system through the existing ESB of the bank)
3. Use data bus and data warehouse to realize data integration . Data ETL processing refers to the collection, transformation, cleaning, storage, integration, and processing of data from hundreds of transaction systems such as bank front-end deposits, loans, credit cards, and wealth management, with heterogeneous technologies and inconsistent data standards, for customer analysis, Performance management and data services provide data support, which is exactly the problem to be solved by data bus and data warehouse, and belongs to the category of data integration. The data bus design process is: through the data collector (IBM product), automatically collect data from the source system at the end of each day and encode and convert it into text format, and through the data processor, perform data quality inspection, incremental data generation, and data standard conversion And so on, transfer the data file to the target system through the data transmitter. The data warehouse design process is as follows: receive the data transmitted by the data bus, load it into the data warehouse system (to choose a product with massive data processing capabilities and horizontal expansion of capabilities), and analyze the data according to the topic model (such as customers, products, accounts, transactions, Channels, etc.) to classify, integrate, and store long-term historical data, build a retail CRM data mart on top of it, and process the data according to the data processing rules of this demand, such as association, aggregation, etc., generate data files, transmit and load To the analytical CRM subsystem and data service system, providing data support for customer analysis, performance management, and transaction inquiries. (Data integration using data bus and data warehouse)
After a year of development and testing, the project will be launched in February 2021. So far, the system has been running stably with good results, which has favorably supported retail data analysis and customer management. At the project summary meeting, the leaders highly agreed with the system division method and the interface, application and data integration methods adopted. (Project operation status)
In recent years, the business proposes that the call center customer service agents need to inquire about the customer's historical transaction details. Due to the adoption of ESB application integration technology, it can be easily connected to the call center system. The data dimensions and indicators for customer analysis and performance management often change and need to be recalculated. Since the data warehouse has accumulated long-term historical data, it is convenient to deal with data change requirements. However, in the face of customers’ obvious needs to inquire about the transactions that occurred on the day through the online banking and mobile banking systems, since the current data bus only supports data T+1 timeliness, it is necessary to add a quasi-real-time data collection mechanism to the data bus to accurately collect transaction data. Real-time synchronization to the data service system to improve the timeliness of data services. (insufficient items)