Success Story
Touch LATAM moves from a monolith to microservices on Amazon EKS to scale across Latin America
At a glance
Touch is a marketing services and sales tools company with more than 18 years of experience, 60,000 employees in 12 countries and more than 100 clients. To grow in Latin America, it needed an infrastructure able to scale and handle more traffic. Craftech migrated its monolithic architecture to containerized microservices on Amazon EKS. Today Touch deploys faster, with better availability and lower costs.
Touch is a marketing services and sales tools company with more than 18 years of experience, 60,000 employees in 12 countries and more than 100 clients. It offers point-of-sale execution, marketing consulting, service teams and BTL.
Solution implemented
- Microservices on Amazon EKS with a mix of Spot and On-Demand Instances
- Capacity managed by Karpenter to provision nodes quickly and optimize costs
- Amazon CloudFront as a CDN to serve the website with low latency, with origin access identity so it is only accessed through CloudFront and not directly from S3
- CI/CD pipelines and DevOps practices
- Infrastructure as code with Terraform and automation with Atlantis
- Metrics and logs for every microservice with Prometheus and Thanos, Promtail and Loki, and dashboards in Grafana
- Central EKS cluster for monitoring and alerting
Business Challenge
Touch's goal was to offer its services in more Latin American countries and multiply its customer base. To do that, it needed an infrastructure able to scale and handle more traffic: migrating from a monolithic architecture to containerized microservices on Amazon EKS.
Solution
To scale quickly, Craftech deployed the microservices on Amazon EKS with a mix of Spot and On-Demand Instances, with capacity managed by Karpenter to provision nodes quickly and optimize costs. The website is served with low latency through Amazon CloudFront, with origin access identity so it is only accessed through CloudFront and not directly from S3. To speed up deliveries, it implemented CI/CD pipelines and DevOps practices, with infrastructure as code in Terraform and automation with Atlantis. For observability, it added metrics and logs for every microservice: Prometheus with Thanos to query multiple clusters in a unified way, Promtail and Loki for logs and Grafana for dashboards, with a central EKS cluster for monitoring and alerting.
Outcomes
- Faster deployments, which support the daily release cycle
- Better availability and stability with autoscaling
- Unified monitoring of metrics across multiple Kubernetes clusters with Thanos
- Lower costs, better scalability and reliability since the migration to EKS
Stack & technologies
