The art of automation

For infrastructure , Many services have similar architectures 、 Database interaction and UI, This paper introduces Paypal Through the practice of template automation infrastructure micro service scaffold code , Maximize the time to develop services and ensure the consistency of services . original text ：The Art of Automating Automation^[1]

brief introduction

As PayPal Website reliability and cloud engineering (SRCE, Site Reliability & Cloud Engineering) Part of the team , We are committed to improving efficiency through automation . However , Writing software that allows people to automate their work intuitively is not a simple task . stay PayPal In infrastructure , The automation components of any operation must provide the highest level of reliability 、 Security 、 Efficiency and self-service ability . In this article, we will discuss how to serve in the global network (GNS, Global Network Services) To achieve automation in .

GNS The team has a lot of operations that need to be automated , It ranges from ordinary tasks that require a few hours of human investment , To complex tasks that require the operation and maintenance team to devote many days of energy . For us , The biggest challenge is how to automate these large and small operations without consuming a lot of development cycles , We need to work hard to prove “ROI( Return on investment )” The rationality of , This is particularly important .

Software service requirements

To turn any network function into a service , Need to complete the following series of tasks :

User interface development

• Open to PayPal All services in the development community must have some form of user interface (UI), So that users can use the network like any other service .

Database interaction and table management

• Most services need to design database tables and write database interaction code to manipulate data .

Core business logic

• You need to write and test the actual business logic for a particular service .

And automation tools Terraform Integrate

• Our basic users include the network operation team , They won't use based on UI Service for . contrary , They are based on GitOps Mode operation , This means that all services we develop need to be developed Terraform provider And Terraform Integrate .

Certification and authorization

• This involves standard enterprise authentication ( Such as single sign on 、 Multifactor Authentication 、 Role management, etc ) Integration of .

Service account ,API Consumer throttling and speed limits

• We decided to provide... For everything API. therefore , All services we develop must provide service accounts , And ensure throttling and speed limit .

As you can see from the list above , Building a viable network service requires a lot of time to develop important peripheral components . In an ideal situation , We can only spend time writing core business logic , And don't do anything else . In two ways , We have reached the ideal state of writing less code .

The first way to solve this problem is to migrate to real micro service components and build pluggable Library Architecture , The second method is template based automatic code generation , Can be generated based on a generic template UI、 Back end and database layer code .

Code generation

For a long time , Data modeling has always been the pillar of well-designed software products . Once the basic data building blocks are identified , It's easier to build software components that work around data . If the data is obtained in a standard format , Then there is a lot of automation space for various software components around data .

This is exactly the method we use . All of our team's projects start with identifying the data model . The data model uses YAML Format defined , In this way, our dynamic code generation module can generate code for all parts of the system . The following figure shows the process :

UI Code generation

React The framework has been included PayPal Widely used in the industry including .PayPal The application team has a very modular React library , For building customer facing applications . We build on this library , Use in any suitable place .

Most system programmers are not familiar with UI Programming . In order to promote rapid UI Component development , We rely on Model Driven Design . Use model driven design , Typical GNS Developers only need to specify UI Layout , Don't write anything UI Code .

GNS The code generation engine adopts a user-defined model , And use React frame , Generate a separate React Code container , With fully functional UI. This approach allows developers to focus on the business logic of the application .

Back end code generation

In the past , Development and supply REST API Microservices with functions are very cumbersome , Developers need to write software for each endpoint and configure it web The service gateway . However ,Flask The framework has been adopted as a standard framework Python Widely recognized by developers . Allied ,Golang In addition to built-in support , also gorilla/mux frame .

majority GNS Service is to use Golang or Python Compiling . With the help of the above framework , Designed based on Jinja Templates generate generic code GNS Code generation engine .

Based on YAML Format defined data model , The code generation engine uses predefined Jinja Template to generate the basic code of the service .GNS Developers can implement precise business logic based on these codes , To reduce programming errors , And eliminate the tedious work of writing everything from scratch .

Database layer code generation

majority GNS Services need some form of database to store the information described in the modeling phase . Write code to add 、 Deleting and modifying entries in the database is cumbersome , And it's prone to bug、 Inconsistent style, etc .Python and Golang Have the ability to handle object relational mapping (ORM, Object Relationship Mapper ) Open source library , We choose to adopt sqlalchemy To standardize database operations .

Because the model is made of YAML Format defined , Our code generation engine not only generates back-end code , Also generate database processing methods .

Terraform Provider Code generation

Terraform It has become the most commonly used tool for automation infrastructure . As GNS Service provider , We have decided that all automation services will be targeted at Terraform The interface of , Thus, it can adapt to the continuous deployment pipeline being adopted .

Terraform The architecture is flexible and easy to expand , Be able to add... To any automation we like Provider. see also terrform Documentation for more details .

because Terraform Provider Is a well constructed library , Our code generation engine can be based on Jinja Template generation is based on golang Of Provider Program , Then build the code into a pluggable module , The module can be installed into the running terraform In the example . Once that is done , The user can be in GNS Carry out standard service terraform operation , Like any other service .

Conclusion

Most of the tasks required to automate any web service have been generated by our code generation engine , This allows us to focus only on the core logic of the service . The shift to microservice architecture and the adoption of template based code generation and model driven architecture enable us to produce services faster . At the time of writing , We can complete the working framework of any service in one day , This helps the team focus on the core logic , And generate more services in a shorter time .

References:
[1] The Art of Automating Automation: https://medium.com/paypal-tech/the-art-of-automating-automation-17b32594a41f

Hello , I'm Yu fan , stay Motorola Have done research and development , Now in Mavenir Do technical work , Pair communication 、 The Internet 、 Back end architecture 、 Cloud native 、DevOps、CICD、 Blockchain 、AI And other technologies have always maintained a strong interest , I usually like reading 、 reflection , Believe in continuous learning 、 Lifelong growth , Welcome to exchange and study together .
WeChat official account ：DeepNoMind

- END -