Pulse Sight
Enterprise monitoring tools assume you have a dedicated SRE team. When your engineering team is small, you need something that gives you the full picture without the configuration overhead. Pulse Sight ingests Grafana metrics, aggregates time-series data with automatic rollups, and renders interactive dashboards with RBAC, alerts, and capacity forecasting.
Built for a production GCP platform, Pulse Sight gives engineering teams a single pane of glass for infrastructure health, application performance, cost tracking, and capacity planning.
The system polls a Grafana instance on a configurable schedule, normalizes metrics through a mapping layer, and persists them to a local database with automatic rollup aggregation (1-minute to 5-minute to 1-hour granularity) and retention policies. Grafana's data model assumes you want to query everything in real-time. For a dashboard that people check a few times a day, pre-aggregated rollups are a better tradeoff.
Includes RBAC with three roles (super admin, colleague, intern), OAuth authentication via Microsoft Entra ID, threshold-based email alerts, and region-scoped access control. Deployed via Docker Compose with Terraform-managed GCP infrastructure, Nginx reverse proxy, and SSL termination.
Architecture
Four layers: data ingestion from Grafana via configurable metric mapping, a storage layer with automatic rollup aggregation, a FastAPI backend with 15 route modules and background scheduling, and a React frontend with 45+ widget components.
Dashboard Pages
Seven pages, each focused on a distinct operational concern.
| Page | Purpose |
|---|---|
| Overview | System health at a glance, key metrics, service status |
| Performance | API response times, database queries, BigQuery performance, connection pool health |
| Costs | GCP billing breakdown, cost trends, budget status, SKU analysis, optimization opportunities |
| Alerts | Threshold-based alerts, email notifications, alert history |
| Capacity Planning | Resource forecasting, scaling recommendations, risk assessment |
| Customer Monitoring | Per-customer metrics, data source coverage, activity logs, infrastructure usage |
| Admin | User management, RBAC, region assignments, cache management, scheduler jobs |
Widgets
The widget library contains 45+ specialized components, each handling its own data fetching, loading states, and error boundaries. Key categories:
| Category | Widgets |
|---|---|
| Infrastructure | CPU, memory, disk, network utilization with breakdown views |
| Performance | API endpoints, BigQuery queries, slow query detection, connection pool health |
| Cost | Budget status, cost forecast, hourly heatmap, SKU analyzer, optimization opportunities |
| Customer | Data source coverage map, activity logs, resource breakdown, budget overview |
| Capacity | Forecast projections, scaling recommendations, risk assessment |
| System | Service status, scheduler jobs, data integrity checks |