DC/OS is based on the production proven Apache Mesos distributed systems kernel, combining years of real-life experience with best practices for building and running modern applications in production.
Build modern apps using state of the art technologies such as containers and big data services, and confidently move from development to production.
You can learn more about dc/os from here
Why dc/os in Vimpelcom?
At VimpelCom we are embarking on the journey of transformation and intend to become one of the pioneers of digital disruption.
Most of our microservice applications are running on containers and container orchestration is pretty easy with dc/os. You can also use other tools like docker-swarm, Kubernetes etc for this purpose however there are some apps which are not running in containers & connected to some microservice container applications. So we need way to provide HA, reliability and scalability for those apps as well.
We used AWS for our development/staging environments. Our testing & production env are running on-premises due to data compliance policies.
dc/os provides Cloudformation templates to create the setup in AWS. Click here to see the cloudformation template.
However we cannot use this Cloudformation template for our purpose because our production is running on-premises. So we created an Ansible playbook which can be used in AWS as well as on-premises.
I am not sharing the original source code in here however I am open sourcing a simple ansible playbook which can be used for both in AWS as well as in Datacenter
In order to use this playbook for on-premises, you may need to change some env variables and few alternations which is pretty easy if you’re familiar with Ansible.
You can see the ansible repo from here
FYI, from the below images, you may notice the ELB’s & ip’s, don’t worry 🙄🙄 it’s a temporary cluster created for testing this playbook 😛😜😜
Once you installed, you can see the number of nodes like below
For installing dcos cli
For running a single service
For scaling the running service
Here you can see the service has been scaled to 5