Skip to main content

​
High-Level Components

  • Frontend (Next.js): Dashboard UI, server-side API proxy routes, session cookie handling
  • Backend (Django + Ninja): Auth, app management, ingest API, analytics/query endpoints
  • PostgreSQL: Core relational data (users, apps, API keys, sessions, environments, endpoint metadata)
  • ClickHouse: High-volume request event storage and analytical queries

​
Data Flow

  1. Client/service sends telemetry to POST /ingest/requests with app-scoped API key
  2. Ingest service validates payload and app scope
  3. Request events are written to ClickHouse
  4. Endpoint metadata is auto-discovered/updated from method + path
  5. Dashboard queries analytics endpoints with JWT auth
  6. Backend returns filtered aggregates to frontend

​
Tenant Isolation Model

  • User JWT routes are app-scoped through /{app_slug} and validated against owner
  • API-key ingest routes resolve tenant context from key and app relationship
  • Query filters enforce app IDs in analytics SQL

​
Auth Model

  • Passwordless-first via magic links (/auth/magic-link + /auth/verify)
  • Optional password login (/auth/login) only after password is set
  • Password reset flow starts with magic link reset request and ends with authenticated password set

​
Operational Notes

  • ClickHouse is required for analytics/ingest-backed metrics
  • Frontend filtering is hybrid: backend query + safe short-lived cache for responsive UX
  • Endpoint tables and filters are server-driven for scale