Glasswave Streaming
Re-architected infrastructure that cut hosting costs 34% while handling 5x traffic during peak events.
Industry
Media & Entertainment
Timeline
12 weeks
Team
2 engineers, 1 DevOps
Category
Cloud & DevOps
Where things stood
Glasswave's infrastructure was over-provisioned year-round to survive occasional traffic spikes during live events, which meant they were paying peak-capacity prices for off-peak traffic most of the year.
How we got there
Traffic pattern analysis
Analyzed a year of traffic data to separate genuine peak-event load from baseline usage before touching any infrastructure.
Autoscaling on Kubernetes
Migrated to a Kubernetes setup that scales out automatically for live events and back down afterward, instead of running peak capacity permanently.
Observability rollout
Added monitoring and alerting so capacity issues surface as a dashboard warning, not a viewer-reported outage.
What changed
34%
Reduction in hosting costs
5x
Peak traffic handled during live events
0
Event-related outages since migration
Tech stack
Kubernetes
Scales streaming infrastructure out automatically during live events and back down after.
AWS
Hosts the underlying compute and storage Kubernetes scales across.
Terraform
Defines the cluster and networking as code so scaling changes are reviewable, not manual.
Prometheus/Grafana
Surfaces capacity issues on a dashboard before they become a viewer-facing outage.
Docker
Packages services consistently across the autoscaled cluster.
Node.js
Runs the services that scale in and out with Kubernetes during live events.
