Shulou(Shulou.com)06/02 Report--

I. background

At the end of the year, many mobile products will launch their own annual bills, which will attract the attention of the mass media every year. This year, I had the honor to participate in the development of the annual bill of Yixin Fortune APP. This article will take you to explore the technical architecture and R & D logic behind the annual bill of Yixin Fortune, hoping to give you some inspiration.

Second, the overall structure and implementation process of the front end

The technology stack used in the front-end architecture of Credit Fortune's annual bill includes:

The front-end page is made of H5; the percentage of data loading progress, the technology uses the swiper plug-in and some CSS3 dynamic effects; poster generation uses canvas image synthesis, the poster background and QR code into one. In order to complete the MGM tracking, M1 information is embedded in the QR code. III. Data sources and data processing

This annual bill involves customer dimension, sales dimension, customer label and other data, including activity participation, articles, video browsing and other data. The integration of this part of the data comes from the subject data of the data center. The following is the architecture diagram of the data center:

ODS: data source layer, which stores the data extracted from the business system. The original data in the business system is extracted, washed and transferred into this layer. This layer of data is close to the original data, but not equal to the original data. When the data is loaded, a series of standard operations are carried out, such as de-duplication, denoising, table naming, field naming and so on. DW: the data warehouse layer, which is the main body of the data warehouse, builds the data model of the ODS layer according to the topic, is a strategic set that provides all types of data support for the decision-making process at all levels of the enterprise, and is a common collection of all topics. DM: data Mart layer is a wide table for field comparison generated from a business application, which is used to provide subsequent business queries, OLAP analysis, data distribution, and so on. The data in this layer is mainly generated by light summary layer and detail layer data calculation.

In the architecture of the data center, we have established a "customer-centric" label system. The label system manages the data according to several categories, such as population attribute, value index, geographical index, psychological index and so on. The processing mode of the label mainly comes from the lightweight summary or derivative processing of DW and DM layer data, as well as the product prediction generated by some models. This label system supports 360-degree customer portraits and analysis of key contact points, and provides cross-channel and full-flow customer experience optimization and key contact optimization.

The data of this bill mainly comes from business operations, user management and other source system data, which are structurally stored in the database cluster, and have been connected to the data console, and fall into the corresponding topic domain according to scheduled tasks or real-time data. The billing data is processed by its corresponding subject data, and the front end accesses the data through the interface API.

The "real-time push of sales evaluation messages" and "bill communication SMS sending" in billing requirements are supported by the intelligent operation system, which integrates the creation, execution, management, feedback and iteration of operational activities. It can filter customer groups through user attributes, tags, plans, operations and other data to achieve accurate access to goals and improve key indicators and operational efficiency.

The following is the flow chart of the intelligent operation system to create the operation plan:

Real-time push of sales evaluation messages: this function relies on the wormhole real-time platform to drop the data to the database, then configure the data in the intelligent operation system, and finally push the message to the product terminal through the message center and aurora. Bill propagation SMS delivery: select qualified customers according to business rules, configure SMS templates and other contents in the intelligent operation system, and then call notify to send SMS messages to customers through the SMS platform. IV. Technical backstage

User data comes from the data of Yixin Wealth platform itself, including: basic information, browsing information, participation activities and other data. How to ensure that the data is accurately and efficiently conveyed to the front end is a necessary guarantee for back-end development. The asset platform adopts the technical architecture of spring+jersery+oracle+redis+jetCache. In order to enhance the user experience and speed up the response time, the project adopts the flexible combination of cache, non-relational database and traditional relational database to provide better data support.

We also focus on interface response time when docking annual billing requirements. The annual billing user data includes two tables: user activity data and operation data, in which the operation data is a heavyweight table. In order to reduce IO operations in the database, two ways are adopted to reduce IO time:

Minimize the chance of accessing the asset data table according to the tags provided by the data set; take advantage of the new Stream feature of java8 to put the complex SQL logic into the code for processing.

Stream is not a collection element, it is not a data structure and does not hold data, it is about algorithms and computation, more like an advanced version of Iterator.

In addition, Stream also provides parallel technology, when the order of data within the collection is not concerned, parallel Stream disassembly tasks can be used to speed up the processing process. For example, when doing statistics, you need to summarize sub-products, or other operations.

If the complex code logic is directly implemented in SQL, the code will be very lengthy and the execution efficiency will not be high. The logic of the code is to use parallel flow Stream to classify and summarize the relevant data according to the type, and divide one subcategory into another category according to the business scenario of this requirement.

Using Stream parallel flows instead of SQL logic can speed up execution efficiency and reduce response time. If interested students want to know more about the features of Stream, they can refer to the technical documentation. The application of Stream can make the code logic clearer and improve the speed.

V. Summary

This project is completed by the cooperation of multiple teams. This paper combs the annual bill requirements at the technical level. As the time is in a hurry and the content is not very detailed, I hope it can bring you some development ideas. I also hope that users can really feel our intentions.

Source: trust Wealth Management Technical team

Authors: mi Zhihua, Sun Liqiang, Li Li, Zhao Quanchao

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.