Results at a glance
- 1monitoring platform for servers, databases and apps
- 2database data sources with custom queries
- Alertson critical thresholds and unusual activity
- Logs + tracesLoki and Tempo next to the metrics
The brief
The client wanted one dashboard for real-time visibility: CPU, memory and server health, database performance, and alerts when something unusual happens, built so it could grow with their infrastructure.
What I delivered
- Grafana platform installed and configured on a Debian/Ubuntu server, ready for the client's team to use.
- Database data sources: MySQL and MongoDB connected, with custom queries that turn raw tables into the metrics the business actually watches.
- Dashboards with dynamic panels and variables for CPU, memory, disk and server metrics, and for database performance.
- Alerting on critical thresholds and unusual activity, so the client hears about problems before their users do.
- Loki for log aggregation and Tempo for distributed tracing, so metrics, logs and traces live in the same place.
Challenges
- Accurate real-time data from two databases. Queries had to be efficient enough to refresh often without loading the databases they measure.
- Alerts that mean something. Thresholds were tuned to catch real problems without flooding the client with noise.
- Room to grow. Dashboards use variables and templating, so new servers appear without rebuilding anything.
Outcome
The client went from no central view to one Grafana platform covering infrastructure, databases, logs and traces, with alerts for the conditions that need attention.
