Host Grafana without installing anything
Dashboards laid over the data you already have: Grafana stores no metrics of its own, it queries your databases and time series and draws the curves.
Starting at €4.99/month
Introduction
Grafana installed, updated and backed up by us, on a machine in France. Dashboards and charts built from your own data sources: SQL databases, time series, CSV files. You install nothing and configure no server: the application is online a few minutes after you order, under your own name.
The plan that suits it is App Essentiel at €4.99/month incl. VAT — 2 GB of memory. It is worth knowing where that memory goes, because it does not go into storing your measurements: Grafana keeps none. It goes into running each panel's query when a dashboard opens, assembling the series that come back, and continuously evaluating your alert rules. A wall display refreshing twelve panels every thirty seconds works harder than one might think.
This is the most misunderstood thing about Grafana, and it is also its strength: it connects, it does not collect. Prometheus, InfluxDB, Loki for logs, Elasticsearch, MySQL, PostgreSQL, SQL Server, Graphite — the data stays where it lives and Grafana reads it. Your dashboards, for their part, are JSON: they export, they diff, they restore identically somewhere else.
What Grafana can do
It plugs into what exists
Time series with Prometheus or InfluxDB, logs with Loki, and above all relational databases: a SQL query against your production database becomes a curve, with nothing exported and no data duplicated.
One dashboard for a hundred machines
Variables turn a screen into a template: a "server" or "customer" dropdown re-points every panel at once. You write the dashboard once and vary it without copying it.
Alerts Grafana evaluates itself
A rule watches a query and fires when the threshold is crossed. Destinations — email, Slack, Telegram, webhook — are configured separately from the rules, and a silence covers planned work.
Dashboards as JSON, plus a public library
Every screen exports to a readable file, goes under version control and re-imports elsewhere. The community library holds thousands of ready-made dashboards: paste a number, the screen appears, then you adjust it.
What people actually use it for
Seeing your sales without buying a BI tool
An online retailer points Grafana at the shop's MySQL database: orders per day, average basket, products out of stock. One SQL query, one panel, and the screen keeps itself current.
Following the load across a fleet
An agency collects measurements with Prometheus and displays them per client through a variable. A single screen covers thirty servers, and the three-month trend reads at a glance.
Displaying sensor readings
A manufacturing SME records temperature and consumption in InfluxDB. The screen on the workshop wall shows the current day, and the history feeds the end-of-year decisions.
Publishing an open dashboard
A local authority shows the energy consumption of its buildings on a screen anyone can consult. The data stays in its own database; only the view is shared.
Finding an error in the logs
A developer ships application logs into Loki, then lines the error spike up against the load curve over the same time window. The cause becomes obvious far faster than by reading a file.
Hosted in France
In our own infrastructure in Nouvelle-Aquitaine, which we operate ourselves. Your business indicators do not travel through a foreign observability platform.
Frequently asked questions
Nothing for Grafana itself. It does, however, need at least one data source to query: a database you already own works perfectly well as a starting point.
No, and this is the most common misunderstanding. Grafana displays, it does not measure: something has to produce the data — Prometheus with its exporters, InfluxDB fed by an agent, or simply your application database. If your need boils down to "tell me when the site stops answering", Uptime Kuma or Gatus do that on their own and straight away. Grafana becomes useful when you want the trend, not just the incident.
Yes, and it is often the fastest way to start. MySQL, PostgreSQL and SQL Server are first-class sources: a query returning a time column and a value column becomes a curve. Plenty of genuinely useful first dashboards are nothing more than three queries against a business database that was already there.
Not at the start. The public dashboard library holds thousands of shared screens, each referenced by a number: paste that number into the import screen and the whole thing appears, panels and queries included. You start from there and remove what you do not need — quicker than facing a blank page.
Yes, they are part of the service. A self-hosted application left un-updated is a security risk, not a saving.