Challenge
A staking-infrastructure provider operates validators across many networks. Each chain exposes its own view of the numbers that matter — gas costs on L1, oracle-feed behaviour, validator performance on emerging networks — and none of them speak to each other. Operations and finance needed one warehouse where cross-chain questions have answers.
Approach
- Per-chain collectors deployed on Kubernetes, each speaking its network’s native interfaces: L1 gas telemetry, oracle-network scanning, validator status on a new testnet.
- A Go API service in front of one network’s node so downstream consumers get a stable contract instead of chain-version churn.
- BigQuery as the analytical core — per-chain models normalized far enough that costs and performance compare across networks without pretending the chains are identical.
- Kubernetes-native operations — RBAC-scoped access, declarative specs, collectors that restart themselves.
Result
- Ops and finance query one warehouse for cross-chain cost and performance questions.
- New networks onboard as another collector + model pair against an established pattern.
- The platform survived chain upgrades without breaking its consumers — the contract held.