Shulou(Shulou.com)06/03 Report--

With the development of the information age, new technologies, new frameworks and new languages emerge one after another, and the technical perspective of solving problems has never changed. All applications need to be associated with the storage system, whether the storage is SQL or NOSQL. Business systems and databases follow different development specifications. In order to make development easier, there is a kind of framework to help solve the transformation from application layer to database, and the famous ORM class framework is one of them. In fact, the main challenge of data center technology is how to unify computing services and all kinds of data storage conveniently, and how to connect with foreground business layer through service-oriented API.

When we discuss mid-Taiwan applications, it will be easier to understand the nature of design ideas by sorting out some methods, including design and architecture. Architecture deals with flexibility, scalability, availability, security, and other architectural designs that are directly related to the business perspective.

Common architectures are as follows:

▪︎ Serverless Architecture:

The Serverless architecture is an application design that includes third-party back-end as a service (BaaS) services, including custom code that runs as a managed, temporary container on the function as a Service (FaaS) platform.

▪︎ Event-Driven Architecture:

The Event-Driven architecture pattern is based on event-driven generation, detection, consumption, and response.

▪︎ Microservices Architecture:

It is a variant of Service-oriented Architecture (SOA) that builds applications as loosely coupled collections of services. In the micro-service architecture, services are fine-grained and protocols are lightweight.

Mid-Taiwan applications will involve multiple integration with multiple systems, so from the engineering point of view of the program, the ancient Roman strategy should be used: divide and conquer, decompose complexity into small chunks, and be scalable. Free use of implementation methods to achieve the goal results, do not adhere to simple means of implementation.

The challenge here is that application development and data development have different data objects and processing methods. Application development is OOP, functional programming, conventional collection, key-value data, while data development needs to deal with dynamic structural data and complex associated operations repeatedly. Therefore, from the point of view of system structure, we need a converter between application development and data development.

Java8 introduces Lambda expressions and streams, which are very attractive to many data developers, but hard coding takes a long time and a lot of time is wasted due to human factors that can lead to error and repetitive work. In order to make this process easier and reduce the number of errors, it has gradually become mainstream with the help of mature computing framework and DSL language.

For example, take the aggregator as an example, the script is as follows:

AB1 [mysql1,mysql2,mysql3,mysql4] 2fork A1=connect (A2) 3

= B2.query@x ("select product_no,sum (allDuration) sallDuration,sum (allTimes) sallTimes,sum (localDuration) slocalDuration, sum (localTimes) slocalTimes from userService where I0419? group by product_no", argType) 4=A2.conj ()

5=A4.groups (product_no;sum (sallDuration): ad,sum (sallTimes): at,sum (slocalDuration): ld,sum (slocalTimes): lt)

A telecom enterprise uses a database table userService to store user service information. Test0 query needs to show the call time, number of calls, local dialing time and local dialing times of all kinds of telecommunications products. In practical use, it is found that the amount of data is too large and the query efficiency is very low, resulting in the front-end display is very slow.

The intermediate computing layer is introduced to merge and summarize the userService data located in multiple databases. Multithreaded parallelism not only greatly improves performance, it is also easier to implement with SPL scripts, and it is also convenient for Java to call SPL. If the above calculation actions are hard-coded, multithreading, data merging and secondary summary, the workload will be very huge.

Using Zhongtai computing components is a good strategy, it adapts to all kinds of SQL, NOSQL and big data platforms, uses consistent structured data model / syntax for development, provides common interfaces and standard result sets, and applications can continue to encapsulate and provide services according to their own needs. The central computing component had better have the data caching feature, which will not have a direct impact on the storage system when a large number of visits are made. From a maintenance perspective, it can easily modify computational logic without affecting other code, constantly optimize algorithms and take advantage of caching, which is the most desirable way for other developers and storage system maintainers to see. But by adding a layer to the development side, it destroys simplicity.

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.