60 minutes, on-site development and delivery of small programs + Web

Shenzhen Airport Project

The first demonstration is the project of Shenzhen Airport. If you have taken a plane in Shenzhen Airport, you should have used this application (you can use wechat to scan the following small program code to experience). The application layer of the whole Shenzhen Airport project is built on the cloud development, including small program and Web. The small program side communicates with the airport application layer deployed on the cloud development through encrypted communication. The application layer mainly includes cloud functions, wechat cloud invocation, etc., bearing desensitization, assembly, monitoring, log, operation logic and business logic for docking with wechat. With the help of the Serverless framework of cloud development, the traditional server purchase and operation and maintenance can be directly entered into the development, saving operation and maintenance manpower. By using the cloud to develop the built-in log system, access to user login status, mobile phone number and other native wechat cloud invocation capabilities, help the business to greatly improve the development efficiency, take the login status, access to mobile phone number function, for example, can shorten from 1 day to 5 minutes. The log system does not need to be built and used, saving at least 5 days. On the Web side, the JS SDK developed in the cloud connects to the application layer and manages various services.

More Case

As of October 2019, cloud development service has more than 500,000 developers, including a large number of head customers such as Pinduoduo, wechat Reading, Tencent News, Weimeng and wechat Pay. WeChat read small program, for example, “a team for life infinite card”, “good books turn out gifts”, “reading group”, “the answer exactly” etc. Most of the functions are based on the cloud and game development to build, use the function of cloud and cloud storage, cloud database capacity, dispense the backend development link, front students into the whole development of stack, can focus on business code, They achieved a total of 349 releases in 10 months, greatly improving efficiency. In addition, not only the small program, there are many developers based on the cloud development of their own Web, Flutter and other multi-terminal mobile applications **, the specific case will not be introduced, if you are interested in the following cloud development small code to learn more about the use of gestures.

Develop the backend using cloud development

1. New database collection HackerNews
2. Develop the HackerNews API cloud function, as shown in the following figure
3. One-click deployment goes online
   • @ cloudbase/cli tool docs.cloudbase.net/cli/intro.h…
   • VS code plug-in docs.cloudbase.net/vscode/intr…
   • Wechat developer tools

Web access

Next we will put the previous small program one click to support Web access. There is no need for additional development on the front end, since Taro is a cross-end framework, and parts of its interface can be compiled into h5 patterns.



The only change we need to make is to adapt the method of calling cloud development. The specific adaptation code is shown in the screenshot below



Here we introduce the JSSDK for cloud developmentGithub.com/TencentClou…. On the Web side we need to log in to call cloud functions. For the sake of demonstration, we use anonymous login. Cloud development supports many login methods. We can open the console of Tencent Cloud and see wechat public account, wechat open platform, anonymous login and more open custom login methods in the login authorization.

Install a CMS extension with one click

The next step is to demonstrate an ability that doesn’t require developing code, based on our extended ability to install a CMS system with the click of a button to manage the application we just built.









We can deploy such a CMS content management system through the extension management interface in the background of cloud development. For the specific installation tutorial, please refer to the official documents:Cloud.tencent.com/document/pr….

Here’s how the CMS extension works,The essence of extension is the application logic plus cloud functions, cloud databases, cloud storage and other cloud resources, which can be understood as a micro full stack of application orchestration.





After the CMS is deployed with one click, we can manage the data of our entire application based on this system, no additional development work is required, and administrator/operator authentication is provided.

Dive In Deep

The previous demonstration gave you an intuitive understanding and understanding of the capabilities of the CloudBase platform for cloud development. The following will lead you to in-depth understanding of the CloudBase platform and cloud developmentHow to enable half a million developers to develop and deploy applications quickly, without caring about operation and maintenance, and get high performance, stable quality and secure online services?

Cloud development overall product architecture

The overall product architecture of cloud development is shown in the figure. On top of the underlying basic capabilities, cloud development also provides a variety of back-end service capabilities and extension capabilities of cloud development, as well as multi-end SDK products and development tools to facilitate multi-end application access to cloud development.

Cloud development engineering capabilities

How does the cloud development platform enable 500, 000 developers to develop and deploy applications quickly, without having to worry about operations and expansion, to obtain high performance, stable quality and secure online services? It is not enough to solve the basic ability of application development, but the engineering capacity building of cloud development platform is the main benefit. The construction of cloud development engineering capacity mainly includes the following aspects:

• The efficiency of
• The quality of
• security
• Multiterminal construction

Quality of engineering capability

In addition to providing these efficient development methods, it is important to consider how our application goes from Demo to production. There are performance and quality issues to consider. How can we ensure the quality of our application’s performance without the developer having to care about the server? Resource and performance guarantee Firstly, resource and performance guarantee ** Cloud development can provide a large resource pool, which can be automatically expanded in seconds during the peak of service traffic, ensuring high service availability. For example, in the aspect of cloud function, cloud development currently provides developers with the ability to support a single cloud function with 1000 concurrent functions. Assuming that the execution time of cloud function is 50ms, a single cloud function can reach 20,000 QPS, which can meet the needs of most user scenarios. The total QPS of 50 cloud functions will reach 1 million QPS. The developer doesn’t have to worry about the machine, the resource automatically scales according to the number of requests, and the developer doesn’t have to pay for idle resources.

To date, cloud development is called more than 700 million times a day.







Technical architecture and optimization of the cloud development data plane

We’re going to take a look at thatTechnical architecture and optimization of the cloud development data planeAs can be seen from the figure, a service invocation by a user needs to go through multiple middle-tier services and finally reach the customer resource layer. The high availability and high performance of cloud development services can be divided into two aspects: “data link” and “underlying resources” on a large level. At present, the availability of “data links” and “underlying resources” of the cloud development system is more than 99.99%.



The data link provides the connectivity between the user side and the cloud resource side. The data link not only carries out data forwarding and transmission, but also includes a series of processing logic. Such as wechat login authentication, ticket generation and verification, data codec, cloud account identity authentication, cluster routing, cloud resource information binding, security rules, data signature and verification, context information injection, access count, concurrency control and other ability support modules.





Availability and performance tuning of cloud functions



Cloud functions are also heavily optimized in terms of performance. Developers familiar with cloud development should be aware that a function can be started in a single call, which can be divided into cold start and hot start. Hot start is very fast, while cold start is relatively slow.



The process of cold start is shown in the figure. By comparison, it can be seen that cold start requires the creation of function instances, and the creation process includes many sub-tasks, such as downloading function code and deploying functions, which usually takes several hundred milliseconds.



Currently, many optimizations have been made on the function side:

• Optimize the code download time, establish multi-level cache in the VIRTUAL machine and availability zone, and improve the download speed of cloud function code package;
• Optimize function network deployment policies to reduce deployment time.
• Optimize container startup and reduce startup time;
• Based on the changing trend of user request volume and the changing rule of periodic request volume, the quasi-real-time prediction of cloud function concurrent volume was carried out, and the capacity was expanded and contracted in advance.
• Support version alias, users Update code, with Rolling Update method to replace the old and new versions, to ensure hot start rate;
• Optimize the reserve time of function instances to ensure the overall low cold start rate.
• The cloud function also uses the lightweight virtualization technology developed by Tencent Cloud, which shortens the startup time of MicroVm to 90 ms and the cold startup time of function to 200 ms, and supports the simultaneous expansion of tens of thousands of compute nodes.

At present, through the optimization of function base architecture and performance, the time of deploying cloud function instances has been reduced, and the cold start time of cloud function has been reduced. Provide customers with higher availability and performance support.





High availability optimization for cloud databases



As can be seen from the architecture diagram, the cloud database access layer is designed in layers and supports horizontal expansion. All user requests are allocated to different hosts, and each host independently maintains user database connections. This ensures that the database access layer can be expanded in a large scale. In addition, multiple clusters are deployed so that faulty hosts can be automatically detected and removed from the cluster. Different users are assigned to different access clusters. User requests can be flexibly scheduled in multiple clusters to cope with possible service faults, providing higher availability and shorter recovery time.

• The publishing strategy can also be more reliable. The gray scale is carried out according to the cluster, and the impact range is small.
• The database engine instances are also deployed across hosts with multiple database instances, which can be automatically transferred when a single database instance fails.
• Data is fully backed up for each data instance every day and saved to the object storage to ensure user data security. When the user data becomes faulty, the data can be reversed, which indirectly improves service availability for users.
• The data layer also supports online live migration. When the host load is too high, user instances are automatically live migrated online to the host with low load. Users are almost unaware of this process. Live migration also supports global load balancing between database hosts.
• Database currently has capability of providing automatic indexing, automatic analysis of user query requests and targeted for automatic index optimization, to users without awareness to optimize query performance, so developers can less concerned with database performance optimization can have excellent performance experience, reduce the burden of the customer, focus more on its own business logic.

Engineering capacity

Finally, cloud development has built-in support for various client SDKS such as wechat applet, QQ applet, Web, Flutter, C#, etc., as well as server SDKS such as node.js, PHP/ Java, etc. The community also provides a large number of SDKS, the full list can be seenwww.cloudbase.net/sdk.htmlPage. End-to-end applications and cloud services are seamlessly connected, and the invocation of cloud services becomes as natural and simple as local ones, greatly reducing the cost of developers to develop multi-end and cross-end applications.

Front-end history

In terms of the development history of the front end, let’s go back to the prehistory of the front end, or the cutting diagram era, where the front end students were responsible for the logic of the view layer, and the rest of the rendering, HTTP API, and business logic were all handled by the back end students.







Then enter the front-end engineering era, this era began to pay great attention to the performance optimization of the front-end, the rendering of the page gradually handed over to the front-end students, the front and back end gradually began to separate.







Then, with the emergence of Node, some front-end students began to take charge of the development of BFF access layer, gradually began to take charge of the full stack development of some innovative projects, and provided HTTP API for external applications.

Serverless application center solution

Serverless provides a new solution for front-end development students. With Serverless, the underlying platform layer assumes the work of the infrastructure. Many underlying technologies and capabilities are gradually sinking, and developers no longer need to pay attention to the operation and expansion of the infrastructure.With a strong underlying architecture, front-end students can quickly realize business development and help businesses realize business value quickly.



In addition to the realization of business value, front-end students can also reap new technical value in this round of changeFor example, by sinking some common business capabilities, orchestrating and delivering them through the Serverless architecture, enterprises and teams can reuse these full-stack capabilities, gradually improving enterprise and team application development capabilities, graduallyPrecipitation out of the enterprise and team’s own application platform.