Skip to main content

Observability - Grafana Stack

Datadog charges per host. New Relic charges per ingestion. Both send your metrics and logs to the US. Bunker hosts a complete Grafana stack - Grafana, Prometheus, Loki, Alertmanager - fully managed. Your observability data stays in Europe, and your bill stays readable.

Why the Grafana Stack?​

Grafana is the most widely used open source visualization platform in the world. Paired with Prometheus for metrics and Loki for logs, it forms a complete observability stack that competes head-on with proprietary solutions.

What sets it apart:

  • Open source: Grafana (Apache 2.0), Prometheus (Apache 2.0), Loki (AGPLv3). Auditable code, no black box
  • Metrics with Prometheus: collection, storage, and PromQL queries across all your application and infrastructure metrics
  • Logs with Loki: lightweight and efficient log aggregation, built for Kubernetes. No more wrestling with Elasticsearch clusters
  • Alerts with Alertmanager: smart alert routing to Slack, email, PagerDuty, or OpsGenie
  • Pre-configured dashboards: Kubernetes monitoring ready to go from day one

Comparison with Alternatives​

CriteriaBunkerDatadogNew RelicElastic StackUptrace
Open sourceYes (Apache 2.0)NoNoPartially (SSPL)Yes (BSL)
Sovereign hostingYes (Europe)No (USA)No (USA)Depends on youDepends on you
MetricsPrometheus (PromQL)ProprietaryProprietaryElasticsearchClickHouse
LogsLoki (LogQL)ProprietaryProprietaryElasticsearchClickHouse
AlertingAlertmanagerProprietaryProprietaryWatcherBuilt-in
Custom dashboardsUnlimitedUnlimitedUnlimitedKibanaBuilt-in
REST APIYesYesYesYesYes
Predictable costEUR 30 excl. tax/monthPer host + ingestionPer GB ingestedServer + timeSubscription
Ops costZeroZeroZeroTime + serverTime + server

What Really Changes​

Datadog billing is notoriously unpredictable. A traffic spike on a Friday night, and your monthly bill doubles. With Bunker, pricing is fixed: EUR 30 excl. tax/month for the entire stack.

Prometheus and Loki use open query languages (PromQL and LogQL). No vendor lock-in on your metrics or log formats. Grafana dashboards export to JSON and re-import anywhere.

Your data never leaves Europe. For regulated industries (healthcare, finance, public sector), this is not a bonus - it is a requirement.

You Keep Control​

  • JSON export: all dashboards exportable via the UI or REST API
  • Standard formats: Prometheus metrics are portable to any Prometheus-compatible system
  • Log portability: standard log formats, migrate to any collector
  • Migration: if you leave Bunker, reinstall Grafana + Prometheus + Loki on your own cluster and re-import your dashboards
  • Open source licenses: the code stays free, the projects cannot become proprietary

Why Bunker Instead of Self-Hosting?​

You could install the Grafana stack yourself. But here is what that actually looks like:

AspectSelf-hostingBunker
Metrics retentionTo configure30 days included
Log retentionTo configure15 days included
Grafana databasePostgreSQL to manageManaged CNPG
Kubernetes dashboardsTo importPre-configured
AlertmanagerTo configurePre-configured
High availabilityYour responsibilityIncluded (SLA target)
BackupsTo configureAutomatic
Security updatesTo monitorApplied
NetworkPoliciesTo configureApplied
24/7 monitoringIronic, right?Included

The critical point with self-hosting is monitoring the monitoring. Who watches Prometheus when Prometheus goes down? At Bunker, the observability stack itself is supervised by an independent monitoring loop.

Quick Start​

1. Deploy from the Console​

  1. Go to console.getbunker.net
  2. Create your account or log in
  3. Deploy the Grafana Stack from the service catalog
  4. A few minutes later, Grafana, Prometheus, Loki and Alertmanager are ready

2. Access Grafana​

Once the deployment is complete, access your Grafana instance via the URL provided in the console. Admin credentials are available in the service details.

3. First Dashboards​

Kubernetes dashboards come pre-configured. Right after deployment, you can explore:

  • Kubernetes / Cluster: CPU, memory, and network overview for the cluster
  • Kubernetes / Pods: per-pod consumption, restarts, statuses
  • Prometheus / Targets: health status of monitored endpoints
  • Loki / Logs: search and filter your application logs

Advanced Features​

Adding a Data Source​

Grafana natively supports many data sources beyond Prometheus and Loki. To add one, head to Grafana settings:

  1. Go to Configuration > Data Sources
  2. Click Add data source
  3. Choose from available sources: InfluxDB, PostgreSQL, MySQL, Elasticsearch, and many more

Creating a Custom Dashboard​

To build a dashboard tailored to your needs:

  1. Click + > New Dashboard
  2. Add a panel with Add visualization
  3. Select your data source (Prometheus for metrics, Loki for logs)
  4. Write your PromQL or LogQL query
  5. Save

Here is a PromQL query example to track the HTTP 5xx error rate across your services:

sum(rate(http_requests_total{status=~"5.."}[5m]))
  /
sum(rate(http_requests_total[5m]))

Configuring Alerts​

Alertmanager comes pre-configured. To create an alert, you can use the Grafana interface:

  1. Open an existing dashboard panel
  2. Click Alert > Create alert rule
  3. Define the condition (for example: CPU > 80% for 5 minutes)
  4. Configure the notification channel (Slack, email, PagerDuty, OpsGenie)
  5. Save

Grafana REST API​

All dashboards are manageable via the REST API. Here is how to export a dashboard:

# Retrieve a dashboard by its UID
curl -H "Authorization: Bearer YOUR_API_TOKEN" \
  "https://your-instance.getbunker.net/api/dashboards/uid/MY_DASHBOARD_UID"

And to import a dashboard from the Grafana community:

# Import a JSON dashboard
curl -X POST \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d @dashboard.json \
  "https://your-instance.getbunker.net/api/dashboards/db"

Sending Application Logs to Loki​

To send your application logs to Loki, use Promtail or the native HTTP client. Here is a quick test with the curl command:

# Send a test log entry to Loki
curl -X POST \
  -H "Content-Type: application/json" \
  "https://your-loki.getbunker.net/loki/api/v1/push" \
  -d '{
    "streams": [{
      "stream": { "app": "my-service", "env": "production" },
      "values": [["'"$(date +%s)000000000"'", "Service started OK"]]
    }]
  }'

In production, Promtail (or Grafana Agent) automatically collects logs from your Kubernetes pods without manual configuration.

Best Practices​

  • Use consistent Prometheus labels (app, env, team) to simplify PromQL queries
  • Keep label cardinality low: too many distinct values degrades Prometheus performance
  • Structure your logs as JSON to fully leverage LogQL in Loki
  • Create alerts on symptoms (latency, error rate) rather than causes (CPU, memory)
  • Export your dashboards to JSON and version them in Git
  • Use Grafana folders to organize dashboards by team or service

Next Steps​