Bankmonitor

The challenge

“Kubernetes on AWS” “Kubernetes on AWS” optimization of Kops and EC2 based system by migrating to Amazon EKS and retaining CICD system. Automatic, system-wide monitoring and alarm system with cost-effective AWS native solution.

The solution

A long-running Kops-based Kubernetes system had to be migrated to Amazon EKS, which TC2 knew would benefit the customer in several ways because it offers an AWS-managed controlplane, plenty of automation, minor version auto-updates, native security-friendly solutions, and AWS-managed worker nodes option,, which is a huge relief for DevOps experts.

After the preparations, the migration of the old and the new system followed a well-thought-out schedule, during which we regularly consulted with the client about the operation of their application. We carried out this process taking into account the methodology, recommendations and best practices provided by AWS. We start each migration project by developing a test system.

Once the empty EKS system was available, we began moving and testing the functionally essential and critical elements. The application component was moved by extending the existing CD system so that this component could also be tested. Fortunately, this required a minor modification due to the integration of AWS CodeDeploy, which was previously used as a CICD element.

After that, we built the entire application step by step, as well as Route53, Application LoadBalancer, Amazon RDS MySQL elements that can be tightly connected to the application, eliminating old and possibly wasteful implementations.

Thanks to the migration, there was also a change of EC2 generation, which meant a significant cost reduction for the customer.

We used the CloudWatch service to examine the system at the infrastructure level and to provide an alarm solution. For the supervision of the EKS system, the so-called We introduced a native AWS solution called Container Insights. The frontend WordPress system and the RDS databases behind the application were created with a script implemented with a TC2 self-developed managed service (TC2MS) that generates resource tag and event-based CloudWatch alarms. With this system, employees can be notified in both email and Slack when a particular resource has entered an alarm state.