Business case for using self-hosted Grafana for visualisation & alerting while retaining Azure Monitor for ingestion?

Grafana
1

We currently use Azure Monitor (Log Analytics + Metrics) for monitoring and alerting across our Azure estate.

We’re considering a change in architecture:

  • Retain Azure Monitor for:

    • Log ingestion (Log Analytics workspace)

    • Metrics ingestion

    • Data storage & retention

  • Introduce self-hosted Grafana for:

    • Dashboards / visualisation

    • Alerting

    • Jira integration

This is driven by dashboard flexibility and alerting requirements, but we need to justify the operational complexity.

Proposed Architecture

  • Azure Monitor continues ingesting logs/metrics

  • Grafana connects via Azure Monitor / Log Analytics data source

  • Grafana self-hosted (not Grafana Cloud)

  • Multiple Grafana instances behind a Load Balancer

  • Shared database backend for HA

  • 5-second dashboard refresh requirement

Key Questions

  1. What are the practical pros/cons of using Grafana for visualisation while keeping Azure Monitor for ingestion?

  2. What advantages does Grafana Alerting provide over Azure Monitor Alerts in real-world use?

  3. Is self-hosting Grafana (HA behind LB) worth the operational overhead vs staying fully in Azure Monitor?

  4. How mature is Jira integration in Grafana vs Azure Monitor Action Groups?

  5. Are there hidden performance or cost considerations when querying Log Analytics from Grafana at 5-second refresh intervals?

Known Constraints

  • Grafana must be self-hosted (cost reasons — no SaaS).

  • High availability required.

  • Need Jira ticket creation from alerts.

  • Need rich dashboards beyond what Azure Workbooks currently provide.

What I’m Looking For

I’m not looking for marketing comparisons — I need architectural tradeoffs, operational realities, and real-world experience from teams that have:

  • Replaced Azure dashboards with Grafana

  • Or run both side-by-side

  • Or reverted back to Azure Monitor

Especially interested in:

  • Alert noise handling differences

  • Multi-dimensional alerting

  • RBAC complexity

  • Maintenance overhead

  • Scaling implications

If anyone has implemented this pattern (Azure Monitor ingestion + Grafana visualisation/alerting), I’d appreciate insight into whether the complexity is justified.

2

I can answer questions 1, 2 and 5 based on my experience at the company where I work here in Brazil… I work as an Architect with Development teams.

  1. What are the practical pros/cons of using Grafana for visualisation while keeping Azure Monitor for ingestion?

The possibility of creating dashboards that unify views from various cloud providers (Azure is our main one, but we use other alternatives, in addition to our own data center).

  1. What advantages does Grafana Alerting provide over Azure Monitor Alerts in real-world use?

Lower cost compared to using Azure, plus unified alert management because we work with other clouds.

  1. Are there hidden performance or cost considerations when querying Log Analytics from Grafana at 5-second refresh intervals?

In our case, cost reduction is a major factor and leads to increased use of Grafana. Being a healthcare company operating throughout Brazil, we have a considerable volume of database transactions, API requests…