Devops learning notes (II)

I learned from the last article DevOps Of the four knowledge systems “ Agile management ”(Scrum frame )、“ Lean management ”（ Lean Kanban ）, Continue today “ Continuous delivery ” Learning from .

First, let's understand the definition of continuous delivery .

Continuous delivery （ English ：Continuousdelivery, Abbreviation for CD）, It's a software engineering technique , Allow the software product production process to be completed in a short period of time , To ensure that the software can be stable 、 Keep it in a state that can be released at any time . Its goal is to build software 、 Testing and publishing becomes faster and more frequent , This approach can reduce the cost and time of software development , Reduce the risk . —— Excerpt from Baidu Encyclopedia

From this definition , We can see three characteristics of continuous delivery ： Short production cycle 、 Stable release 、 Continuous release .

From the extension of these three characteristics , The whole process of continuous delivery is very similar to the concept of manufacturing production line . To reduce production costs in manufacturing , Increase product output rate , Improve product quality , Formed the concept of manufacturing line . How did the manufacturing line come into being ？

20 At the beginning of the century , American Henry . Ford first adopted the assembly line production method . In his factory , Specialized division of labor is very detailed , Only one production unit has as many processes as 7882 Kind of , In order to improve the labor efficiency of workers , Ford tried and tried , Identify the workers needed for an assembly line , And the distance between each process . thus , The assembly time of each car chassis ranges from 12 Hours 28 The minutes are shortened to 1 Hours 33 branch , And throughout the production line , Can grasp the product quality status in real time , The risk of bad products is greatly reduced .

Return to the information production process , In the past, the process of program development in waterfall management projects started from requirements planning → Software code programming → unit testing → Integration testing → Development environment deployment → Code release → Operation and maintenance , The period is generally calculated on a monthly basis , And it can only be delivered to users after all development tests are completed , The middle time span is large , Business requirements change quickly , Often the final product can not fully meet the needs of customers , This leads to the failure of many projects .

If the production line is combined with software delivery , There is a continuous delivery pipeline , That is to say CI/CD/CO（ Continuous integration - Continuous delivery - Continuous operation ）.

In order to continuously release, in addition to the delivery pipeline , Another important thing is automation , Including the automatic construction of program code 、 Automatic function test 、 Automated non functional testing 、 Automated deployment, etc . Because the delivery cycle is shortened from year to quarter , And then monthly , Another week , Even to the day of delivery , It means that the original code is submitted 、 test 、 The frequency of deployment should also be increased to days . It turns out that only the deployment before the program goes online , Do it every day , It is bound to require a lot of automatic upgrading of work , Only in this way can there be the feasibility of continuous delivery .

According to the concept of continuous delivery , To achieve full automation from code to operation , The corresponding system can be used to manage and realize . Here we introduce the software based on open source platform ,Git： Code version control ,GitLab： Code warehouse ,Maven：Java Language construction ;Nexus : Product warehouse ,JUnit : unit testing ,Sonarqube : Code quality scanning ,Docker : Containers ,Harbor :docker Mirror warehouse ,Jenkins : Process planning ,Kubernetes : Service deployment .

In the process of continuous delivery , Through local development 、 Integrated development 、 test 、 pre-release 、 Continuous release of formal environment and rapid feedback . Improve code quality through continuous feedback .

After the local developer completes the unit test , Submit code to the integration environment , You will receive the result of integration feedback ; No problem with program integration , Continue to commit back to the test environment , Start testing functionality and non functionality , Test defects will be directly fed back to the developers for processing ; After passing the test, continue to submit to the production environment , After deployment , Through the user's use of feedback program operation defects and system optimization requirements , Accordingly, the whole delivery process is traversed again , After tuning, release and launch , Achieve program code quality management .

After learning about continuous delivery , How to build and deploy the pipeline of continuous delivery ？ First of all , Model the value stream and create a simple working framework ; second , Automate the build and deployment process , This is also the most obvious step , Automating build and deployment can significantly reduce the cycle of program release ; Third , Automate unit testing and code analysis , It is helpful to carry out quality analysis of program code and the whole system ; Fourth , Automate acceptance testing ; The fifth , Automate publishing .

---