bespinian supports TX Group in modernization of 20 Minuten application


"The collaboration with bespinian has always been very good. Thanks to their expertise, bespinian made a significant contribution to the successful migration of the digital products from 20 Minuten to a new and modern infrastructure."


Manuel Sutter, Product Owner



Customer

20 Minuten is the media brand with the widest reach in Switzerland. Like the media house Tamedia and the advertising marketer Goldbach, it belongs to the TX Group. 20 Minuten is characterized by its reach in all language regions and the story mix of news, entertainment, and inspiration. The 20 Minuten team was and is an important partner for bespinian in the cooperation regarding the migration of digital products to a new infrastructure, and other projects in the context of OAuth2, cloudification, containerization and modernization of software.


Background

20 Minuten is Switzerland's largest commuter newspaper and one of the most visited websites in Switzerland. It is a media portal with daily news, entertainment and inspirational articles. In 2019, 20 Minuten had planned a major migration of its applications away from the legacy infrastructure and application code, some of which was almost 20 years old. This involved migrating existing VMs to a modern Kubernetes-based container infrastructure on AWS. The backend was to be migrated from Perl to Go and containerized. The frontend was also rendered by Perl and was to be converted into a modern Next.js application. 20 Minuten has over one million users every day. This makes a migration without downtime particularly challenging. In addition, it was important to keep the operating costs within manageable limits despite this large load.


Project Goal

The goal of the project was to modernize the setup of the application. The legacy components and the previous code were to be replaced in order to make the system more stable and cost-efficient. AWS was chosen as the infrastructure provider and the application was to be redeployed in Go, Node.js and React microservices on an EKS (Kubernetes) cluster.


bespinian's Role

bespinian had a key role in the following topics regarding application migration in terms of refactoring / re-architecting from the 6 migration strategies of AWS.


Infrastructure on AWS

bespinian has been instrumental in provisioning the infrastructure on AWS. We played a key role in writing and applying the appropriate Terraform modules and setting up the Kubernetes clusters on which the various microservices are hosted. An important part of this was also the correct configuration of the caching layer (Amazon CloudFront) to keep the costs manageable given the large number of concurrent users of 20 Minuten. bespinian did a lot of the technical implementation and automation, but also the communication between the development and infrastructure teams.


Kubernetes and container configuration

An exciting task was the configuration of the Kubernetes cluster (on Amazon EKS). The architecture as well as the implementation of it still bears bespinian's signature. The main focus of this task was making the necessary architectural decisions and implementing the provisioning and configuration of the cluster via Terraform. Subsequently, another key focus of the task was on automating the deployments of the individual services via Helm and GitHub Actions. bespinian played a key role in all of these areas. We also played a major role in coordinating the individual teams and applications in the sense of a DevOps migration.


Application Development

The various microservices of the backend and the frontend were co-designed by bespinian, both architecturally and in terms of software development. As experienced Go developers, we were able to contribute to the design of the application code itself, but also to its automation, deployment and communication between the services.


Monitoring and Alerting

Monitoring the applications in terms of a DevOps transformation was very important, especially in the initial phase after the migration. A large number of users were migrated to the new applications and sometimes produced access patterns which were difficult to predict. These access patterns first needed to be analyzed and understood in order to provision the appropriate resources and produce the appropriate fixes for performance issues and bugs. Prometheus and Grafana played a major role in this process to get an overview of what is happening and where which requests are being made.


Caching

An important part of bespinian's contribution was the configuration of the CloudFront caching layer. Due to the high number of requests, most of which are read-only, we were able to save very large monthly contributions by clever use of the cache. This allowed the containers to be used in relatively small numbers of replicas. The optimal configuration of the cache with so many different access patterns and clients (20 minute website, different mobile applications and CMS) was a big challenge. bespinian was responsible for tackling this challenge and providing a solution to the 20 Minuten team.


Technologies

  • Amazon Web Services (AWS)
  • Terraform
  • Kubernetes
  • Prometheus / Grafana
  • Helm
  • Go
  • React (Next.js)
  • MongoDB
  • PostgreSQL
  • GitHub Actions