Energy utility, DACH
Migrating a monolithic ERP landscape to the cloud
Zero-downtime migration of 14 legacy services to Kubernetes — with a GitOps pipeline and 70% lower infrastructure costs.
← All projects
Mid-cap fintech
SRE platform for a real-time trading system
Built an SLO framework, observability stack and chaos-engineering practice — mean time to recovery down from 47 to 6 minutes.
At a glance
- Client
- Mid-cap fintech
- Industry
- Financial services
- Service
- Engineering
- Project duration
- 6 months + ongoing operations
- Team
- 2 SREs, 1 engineer
Starting point & goal
A mid-sized fintech runs a real-time trading system where every minute of downtime costs money directly. Incidents were handled reactively and ad hoc — without clear metrics and without dependable alerting.
Challenges
- Mean time to recovery of 47 minutes for critical incidents
- No shared understanding of availability: SLOs were missing
- Alert fatigue from hundreds of unprioritised alerts
- High load peaks at market open
Implementation
We established an SLO framework that makes availability measurable and built an observability stack that shows causes instead of symptoms. Runbooks, on-call processes and regular chaos-engineering exercises keep the team ready to act — before the emergency happens.
Tech stack
Infrastructure
- Kubernetes
- Terraform
DevOps & observability
- Prometheus
- Grafana
- Loki
- PagerDuty
Languages & frameworks
- Go
- TypeScript
Results
01
Mean time to recovery reduced from 47 to 6 minutes
02
99.98% availability in the first year of operations
03
Alert volume cut by 80%