How did we reduce infrastructure costs by around 80%?
Implementation
GCP & AWS
Technology
GKE / EKS
Tooling
Kubernetes, Helm, Golang
Team
3 Engineers
Scale
6000 production containers
BENEFITS
80% Infrastructure cost decrease
Summary:
As we implemented a cloud-native approach throughout the organization, we achieved a level of application resilience that could withstand any hardware failures. This allowed us to take advantage of cheaper spot instances and reduce our infrastructure costs by around 80%. Additionally, we incorporated "chaos engineering" into our daily operations, which proved to be highly effective.
Challenges:
High availability requirements
Over 100 services managed by different teams
Instability of spot market (lack of available servers)
Solution:
Introduce “chaos engineering” events to test applications and infrastructure resilience
PoC implementation for less important projects
Tuning of k8s pod affinity, priorities, auto-scaling strategies