Work / Utility Warehouse
Systems that tell you they are failing
Real time customer services on a container platform, and the observability that made their failures visible.
- Client
- Utility Warehouse
- Sector
- A multi service provider bundling home services into one bill
- Period
- December 2019 to December 2022
The problem
A company that bills for several services at once has a lot of moving parts talking to each other, and a customer who does not care which one broke. Email, chat, calls, reminders and payments all have to work, and a fault in any of them looks the same from outside.
The interesting problem is not building them. It is knowing, at two in the morning, which one is failing and why, without reading five dashboards that were built by different teams for different reasons.
The work was building those services on a container platform and instrumenting them so a failure surfaces before a customer reports it.
What this touches
Worked with
- TypeScript
- React
- GraphQL
- Go
- Kubernetes
- Kafka
- Prometheus
- Grafana
- Terraform
Have the same problem?
Tell us what it is and what is worrying you about it. If it is not something we should take, we will say so.