European Gaming Giant Increases Platform Stability with Kubernetes

Born in 2012, MavenHut rooms more than 40 million players across the world, having launched over 40 games now called “imaginative social experiences”. Just 3 years later, MavenHut would become our first client in the gaming industry and an early adopter of Kubernetes.

Our conversation started with addressing the platform stability at that time—a DigitalOcean and Docker Swarm combo with a layer of scripts to orchestrate the containers. The configuration implied difficulties in managing and deploying stateful applications on Docker Swarm. Long story short, it was not a permanent fix, as scaling was challenging, and deploying new services required assistance. 

Kubernetes was the obvious choice for us. It would provide developers with the flexibility of deploying applications, effortless monitoring of their activity and achieve the required standard of stability and agility. 

The team behind MavenHut was open to the idea of adopting Kubernetes, as they had already assembled a strong development team. Our Cloudhero(s) Andrei Manea, Andrei Petrescu and George Drăghici have teamed up with MavenHut’s Andrei Văduva to craft the migration plan. In just two months, we have performed the actual migration process that consisted of the following steps:

crop-0-0-512-512-0-Andrei-Vaduva.png
Andrei Văduva working on MavenHut
  1. Migrate all Elasticsearch data to Kubernetes
  2. Move all applications running on Docker Swarm to Kubernetes
  3. Create new Jenkins pipelines for Kubernetes
  4. Move Redis to Elasticache
  5. Migrate MySQL data 
  6. Migrate MongoDB
  7. Monitoring using Prometheus and Grafana.

 

Achieving platform stability has been our top priority and Kubernetes provided the foundation for it—you know you’re using the right solution when everything runs smoothly even at 500K daily active users. Below you will find an insight showcasing the number of incidents reported on PagerDuty before and after implementing Kubernetes and AWS. 

Screen Shot 2020-07-04 at 00.12.14
Incidents reported on PagerDuty before Kubernetes & AWS