Cloud Native Zwitscher Showcase (JEE-based)

This showcase demonstrates how to build a cloud native application using JEE and several cloud native infrastructure components. The individual parts will later be deployed and run on DC/OS.

Used technologies

• JEE Service Container: KumuluzEE
• API Gateway / Edge Server: Fabio
• Service Discovery: Consul
• Service Client: Jersey and Hystrix
• Configuration & Coordination: Consul
• Cluster Resource Manager: Mesos by DC/OS
• Cluster / Container Orchestrator: Marathon by DC/OS
• Diagnosability: Metrics

Running the application

The application can be run on your local machine as well as on a remote DC/OS cluster.

Local maschine

• import the gradle project into your favorite IDE. JDK8 required.
• perform a gradle build:

$ ./gradlew clean build

• start the infrastructure (Consul and Fabio):

$ cd infrastructure
$ ./local-consul-up.sh
$ ./local-fabio-up.sh

• pick out a service within the ./services directory and run its Main class. Important: set the PORT environment variable before to free port of your choice.
• now you hopefully see the service running within the JEE container and its service getting registered at Consul and Fabio.

You can also use docker-compose to launch a complete local system (all micro services and all required infrastructure like Consul and Fabio). You need a remote docker daemon like the one provided by docker-machine.

• point your DOCKER_HOST environment variable to the remote docker daemon. If you're using docker-machine on macOS this is as easy as eval "$(docker-machine env default)".
• build your docker files with the buildDockerImage gradle task.
• now you can start the local system with ./infrastructure/dockerized-up.sh and stop it with ./infrastructure/dockerized-down.sh.

DC/OS

• setup a DC/OS Cluster. You can do so locally by using DC/OS Vagrant (https://github.com/dcos/dcos-vagrant) or you use a public cloud provider. As the Vagrant setting consumes a lot of resources our current favorite is the packet (https://www.packet.net) IaaS cloud. It's fast and way cheaper than AWS for running a DC/OS cluster. A DC/OS cluster can be created by a Terraform script on Packet: https://dcos.io/docs/1.8/administration/installing/cloud/packet.
• install the DC/OS command line tool and auth it to your cluster.
• now you can start the cluster with ./infrastructure/marathonized-up.shand stop it with ./infrastructure/marathonized-down.sh.

If you want to deploy the microservices you've to push them to docker hub:

• point your DOCKER_HOST environment variable to the remote docker daemon. If you're using docker-machine on macOS this is as easy as eval "$(docker-machine env default)".
• set your docker hub credentials by using ./gradlew setCredentials --service docker --username <YOUR-DOCKER_HUB-USERNAME>
• Build and deploy the container:

$ ./gradlew clean build buildDockerImage pushDockerImage

License

This software is provided under the MIT open source license, read the LICENSE file for details.