Awesome
Building Scalable CQRS Microservices with Cloud Foundry
This project demonstrates how to build cloud native Command & Query Responsibility Separation and Event Sourcing applications using Pivotal Cloud Foundry.
It combines the following technical elements in order to produce one logical application - a 'Product Data Service'...
- CQRS and Event Sourcing using the Axon CQRS Framework
- Microservices using Spring Boot
- 12-Factor configuration using Spring Cloud Config
- Various other Spring Cloud things in the background like:-
- Spring Cloud Service Registry for service discovery (Netflix Eureka)
- Spring Cloud Bus (for push-enabled configuration changes)
- Spring Cloud Connectors (for connecting to backing-stores like MySQL)
I'm using Pivotal PCF-Dev for this demo.
About PCF-Dev. PCF-Dev is a free open-source 'cloud-foundry-on-your-desktop' application from Pivotal. It's designed for cloud developers; emulates Pivotal Cloud Foundry; and comes with RabbitMQ, MySQL and Spring Cloud Services built-in. Apps tested against PCF-Dev will run against Pivotal Cloud Foundry without modification (assuming that the same backing-services required by your app are present).
If you're not interested in the cloud (why not?) then the original non-PaaS version of this project can be found here: https://github.com/benwilcock/microservice-sampler
If you'd like more general information on CQRS/ES Microservice architecture check out this handy CQRS slideshare.
Getting Started
To follow this tutorial you'll need...
- A Mac, Linux or Windows PC with 16GB RAM
- Java JDK 1.8
- About 15 - 45 minutes depending on your setup.
Install the required software
-
Install the CF CLI command line tool (currently v6.22.1+) as instructed here: https://github.com/cloudfoundry/cli
-
Install VirtualBox 5+ from here: https://www.virtualbox.org
-
Install the PCF-Dev (currently v0.20.0+) developer environment as instructed here: https://github.com/pivotal-cf/pcfdev
Once you've installed these tools, you're ready to create a local development environment that mimics Pivotal Cloud Foundry.
Provision your local CloudFoundry and run the App
- Checkout the source code for this project to your machine and build it:
$ git clone https://github.com/benwilcock/cf-cqrs-microservice-sampler.git
$ cd cf-cqrs-microservice-sampler
$ ./gradlew clean assemble
You don't actually need to have
git
installed if you don't want it, simply download a ZIP of the source-code instead and unzip it.
You don't need to install Gradle. The source code includes a
gradlew.sh
and agradlew.bat
file that you can use to run gradle commands.
- Start PCF-Dev with Spring Cloud Services:
cf dev start -s all
Starting PCF-Dev takes about 8 minutes depending on your PC. It emulates complex cloud infrastructure in a local VirtualBox environemnt. Once started it can be suspended and resumed to save you time.
- Attach to PCF-Dev from the cf CLI:
cf dev target
You have now activated your PCF-Dev client and it's ready to push applications to your local cloud development environment.
- Create a MySQL database backing-service called
mysql
in your local cloud:
cf create-service p-mysql 512mb mysql
- Create a RabbitMQ messaging backing-service called
rabbit
in your local cloud:
cf create-service p-rabbitmq standard rabbit
- Setup the Spring Cloud Config backing-service in your local cloud (takes a few minutes - use
cf services
to check progress):
cf create-service p-config-server standard config -c ./config-server-setup.json
- Create a Spring Cloud Service Registry:
cf create-service p-service-registry standard registry
- Now "Push" the project to your local cloud:
cf push
The push uses the
manifest.yml
file. This file tells the cf CLI where to find the Command and Query side apps and which backing-services to "bind" the apps to.
Check everything works
This demo contains two REST microservices (the command-side and the query-side), so there is no UI. However, you can interact with the services using simple curl
commands as shown below...
- Run the integration tests:
./gradlew integration-test:integrationTest
These simple RESTAssured tests should run and pass.
- Use the command-side to "add" a Product:
curl -X POST http://command.local.pcfdev.io:80/add/fb226b13-65f5-47bf-8a06-f97affaaf60f?name=MyTestProduct
The Commands executed on the command-side should result in events being stored in the MySQL database for
Events
before being sent out via RabbitMQ to the registered message subscribers.
- Use the query-side to view the Product you just added:
curl -X GET http://query.local.pcfdev.io:80/products/fb226b13-65f5-47bf-8a06-f97affaaf60f
The query-side is a registered event message subscriber and is listening to RabbitMQ for these events. When an event is received by the query-side app it then posts records into the
Product
materialised-view (in it's own database). You should see the following JSON output from the query-side app:-
{
"name" : "MyTestProduct",
"saleable" : false,
"_links" : {
"self" : {
"href" : "http://query.local.pcfdev.io/products/fb226b13-65f5-47bf-8a06-f97affaaf60f"
},
"product" : {
"href" : "http://query.local.pcfdev.io/products/fb226b13-65f5-47bf-8a06-f97affaaf60f"
}
}
Notice how the 'commands' go to the command url (command.local.pcfdev.io
) and the queries go to the query url (query.local.pcfdev.io
)s. This means you can scale the command and query apps separately depending on load.
Why use PCF for this App?
There are several clear business benefits that make switching to Cloud Foundry an attractive option for this project.
-
Lower Maintenance
- I don't have to script my own PaaS features (like auto-restart).
- I can provision my own backing-services (like MySQL etc.) without Ops.
- I don't have to create, manage, secure or test container images.
- I can achieve Continuous Delivery and add value constantly.
-
Better Resiliance
- Cloud Foundry will restart my apps for me if they crash.
- Cloud Foundry will scale my apps for me horizontally.
- Cloud Foundry supports zero downtime app upgrades (blue/green).
-
Detailed Operational Insights
- Cloud Foundry includes built in Metrics.
- Cloud Foundry supports external logging services like Splunk and Logstash.
-
Lower Cost and Fewer Resources
- Cloud Foundry uses cloud infrastructure like AWS, Azure, GCP etc.
- Cloud Foundry uses cloud infrastructure efficiently, keeping running costs lows.
- Developers can be more productive and retain greater control.
- Operators can spend less time on the 'scaffolding' and more time on real operational issues.
About the Author
Ben Wilcock works for Pivotal as a Cloud Solutions Architect. Ben has a passion for microservices, cloud and mobile applications and helps Pivotal's Cloud Foundry customers to become more responsive, innovate faster and gain greater returns from their software investments. Ben is also a respected technology blogger who's articles have featured in DZone, Java Code Geeks, InfoQ and more.
Thanks
- Thanks to Markus Bukowski for helping sort out the trixy configuration required to get the transaction manager and Axon unit of work to actually work.