devonfw guide

12-factor-app with devon4j

This document mainly focuses on discussing how can you create 12 factor app with devon4j. To know more about this 12 factors you can refer here . Twelve factor is mainly focus on creating cloud native applications. These are the guidelines on what factors you need to consider in different stages of application lifecycle.

Factor Action

Factor	Action
Codebase	One codebase tracked in revision control, many deploys One codebase should be created with each process or service. Twelve factor app says do not share code between application. If you must share code, build library and manage that through dependencies. To create and manage library in spring boot refer here To build your extension in Quarkus refer here For complex monolith application you can consider breaking it into microservices. One of the good article for it is https://martinfowler.com/articles/break-monolith-into-microservices.html This factor mainly points at having seperate commit history for each service or process in case you are using any submodules consider using submodules in your revision control system as well.
Dependencies	Explicitly declare and isolate dependencies In devon4j we can declare dependencies in pom.xml.Do not commit any jar files into code repository i.e do not commit agent-jars into repository instead consider using maven-dependency-plugin to manage it. And to isolate app from environment we can dockerize devon4j application. References: Dockerize Spring-boot app refer here. Dockerize Quarkus app refer here
Config	Store config in the environment Here, configuration means anything which can vary between different deployment such as creadential to external sources, backing services etc. To do such configuration in your java application refere guide here
Backing services	Treat backing services as attached resources To design your services, refer devon4j guide here
Build, release, run	Strictly separate build and run stages Refer guide here for more details.
Processes	Do not store your states in in-memory cache instead use Redis or Memcached Refer guide for stateless programming in devon4j here
Port binding	The web app exports HTTP as a service by binding to a port, and listening to requests coming in on that port. This is achieved in spring boot application as well as Quarkus. Both have embedded tomcat in it by default so we do not need to deploy explicitly and can be run as standalone application.
Concurrency	As process in application are stateless it will be easy to scale out application based on requirement. Refer guide for stateless programming here because of stateless processes we can scale out easily.
Disposability	Quarkus is designed with container first approach in mind so it has very quick start up time. Comparatively spring boot application use more memory and have slow start-up. Check if start time of your application is less than or equal to minute or you may want to investigate cause for slower start up of your application. To improve your start up time you can refer here For graceful shutdown of web server in spring boot refer document here For graceful shutdown in Quarkus refer document here
Dev/prod parity	With docker this can be achieved. we should have the right processes like continuous integration and delivery to facilitate bridging this gap further.
Logs	Treat logs as event streams 12 factor app treats log as event stream and do not consider details like log files to store etc. In practice to achieve this logs can be written to stdout and stderr etc and variables like log_level can be defined in environment variable or configuration management can take care of it. In cloud provider such as AWS, services like AWS cloudwatch are available. AWS cloudwatch is a monitoring service and can be integrated easily in almost all other services. Redirecting application logs is possible from services like EC2 and ECS to cloudwatch. Similarly other cloud provider have different services for monitoring. Also project can choose to redirect this log events to `datadog`,`splunk`,`ELK` etc
Admin processes	Create scripts for your admin jobs or consider to create Kubernetes job for this processes.

Codebase

One codebase tracked in revision control, many deploys

One codebase should be created with each process or service. Twelve factor app says do not share code between application.
If you must share code, build library and manage that through dependencies.

To create and manage library in spring boot refer here
To build your extension in Quarkus refer here

For complex monolith application you can consider breaking it into microservices. One of the good article for it is
https://martinfowler.com/articles/break-monolith-into-microservices.html

This factor mainly points at having seperate commit history for each service or process in case you are using any submodules consider using submodules in your revision control system as well.

Dependencies

Explicitly declare and isolate dependencies

In devon4j we can declare dependencies in pom.xml.Do not commit any jar files into code repository i.e do not commit agent-jars into repository instead consider using maven-dependency-plugin to manage it. And to isolate app from environment we can dockerize devon4j application.

References:

Dockerize Spring-boot app refer here.

Dockerize Quarkus app refer here

Config

Store config in the environment

Here, configuration means anything which can vary between different deployment such as creadential to external sources, backing services etc. To do such configuration in your java application refere guide here

Backing services

Treat backing services as attached resources

To design your services, refer devon4j guide here

Build, release, run

Strictly separate build and run stages

Refer guide here for more details.

Processes

Do not store your states in in-memory cache instead use Redis or Memcached Refer guide for stateless programming in devon4j here

Port binding

The web app exports HTTP as a service by binding to a port, and listening to requests coming in on that port. This is achieved in spring boot application as well as Quarkus.
Both have embedded tomcat in it by default so we do not need to deploy explicitly and can be run as standalone application.

Concurrency

As process in application are stateless it will be easy to scale out application based on requirement. Refer guide for stateless programming here because of stateless processes we can scale out easily.

Disposability

Quarkus is designed with container first approach in mind so it has very quick start up time.
Comparatively spring boot application use more memory and have slow start-up. Check if start time of your application is less than or equal to minute or you may want to investigate cause for slower start up of your application. To improve your start up time you can refer here

For graceful shutdown of web server in spring boot refer document here

For graceful shutdown in Quarkus refer document here

Dev/prod parity

With docker this can be achieved. we should have the right processes like continuous integration and delivery to facilitate bridging this gap further.

Logs

Treat logs as event streams

12 factor app treats log as event stream and do not consider details like log files to store etc. In practice to achieve this logs can be written to stdout and stderr etc and variables like log_level can be defined in environment variable or configuration management can take care of it.

In cloud provider such as AWS, services like AWS cloudwatch are available. AWS cloudwatch is a monitoring service and can be integrated easily in almost all other services. Redirecting application logs is possible from services like EC2 and ECS to cloudwatch. Similarly other cloud provider have different services for monitoring.

Also project can choose to redirect this log events to datadog,splunk,ELK etc

Admin processes

Create scripts for your admin jobs or consider to create Kubernetes job for this processes.