Grafana for business metrics (SQL dashboards)

Who each one is for

Grafana is the standard answer for infrastructure and observability dashboards. Metrics, logs and traces, graphed over time, with alerting wired to the same data. It runs as free open source you host yourself, or as Grafana Cloud.

arcodash is a SQL-first tool for business data, hosted only. Dashboards sit next to scheduled checks that watch your tables (orders, signups, payments) and notify you when something looks off. The question it answers is not “is the server healthy” but “is this number right, and did the table stop updating”.

People search for “Grafana for business metrics” because they already run Grafana and wonder if it can double as their KPI tool. Often it can. This page is about when it’s worth using something else.

Where Grafana is the better choice

• Infrastructure and observability. CPU, latency, error rates, logs, traces. This is Grafana’s home turf, and arcodash doesn’t compete there.
• Real-time alerting on metrics and logs. Grafana evaluates rules against live time-series. arcodash checks run on a schedule against SQL queries.
• You want it free. Grafana has a free Cloud tier (3 active users, 14-day retention) and a free self-hosted open source edition. arcodash has a 14-day trial and then a fixed $59 or $300 a month, depending on plan.
• You want to self-host. Grafana open source can run on your infra. arcodash can’t be self-hosted.
• Grafana already covers your team. If a few people look at dashboards and the free tier or your self-hosted install does the job, there is no reason to add a tool.
• Your metrics already live in a time-series store. Don’t move them for the sake of it.

Where arcodash fits

• The data is business SQL. Orders in Postgres, payments in MySQL, events in a warehouse. You write the query, chart it, and put it on a dashboard. Everything starts from SQL.
• Checks on the data itself. A freshness, row count, null rate or duplicate check is a saved query with a threshold on one column. See data quality monitors for the four recipes.
• Baseline monitors. When you don’t know the right threshold, a monitor learns what’s normal for a column from its own history. It needs 7 observations before it starts flagging.
• AI on top of SQL. Ask with AI, Fix with AI, suggested checks, and a “Why did this fire?” analysis attached to alerts. You bring your own Gemini, OpenAI or Anthropic key.
• Unlimited viewers at one price. Grafana bills per active user plus usage, so the bill moves with headcount and usage. arcodash is fixed: the price on the page is the bill. More in flat-price BI.

arcodash does have Prometheus and InfluxDB connectors. They exist so you can pull a time-series into a query, the same as any other source. They aren’t a reason to move your monitoring off Grafana or Alertmanager.

Alerts compared

Grafana Alerting is built around alert rules. A rule runs a query against a data source, evaluates a threshold or condition, and routes the result through contact points. It’s available in open source and in Cloud, and it handles the live, high-frequency, label-heavy world of metrics well.

arcodash alerts (called monitors) attach to a saved query. A monitor reads the first row of the published result, takes one column, and compares it with a value using >, <, >=, <=, == or !=. It notifies when the state changes, and a rearm period throttles repeats. Destinations include email, Slack, Teams, webhook, PagerDuty, Mattermost, Discord, Google Chat and Datadog. The alerts overview has the exact semantics and scheduled refresh covers how often a query runs.

For a business number, say “signups in the last hour fell below 5”, the two work in a similar way. Grafana can do it, if you add a SQL data source and write the rule. The difference is what comes with arcodash out of the box: check recipes, baseline monitors, AI-suggested checks and incident tracking. For a CPU spike on a host, use Grafana.

Migrating from Grafana

There is no import tool.

• What ports: any panel that already runs SQL against a SQL data source. Copy the query, check the dialect, save it.
• What you rewrite: PromQL and other metrics queries are not translated. If the data also exists in a SQL database, write the SQL; if it only exists as time-series, leave it in Grafana.
• What you rebuild: dashboards, variables and alert rules. Grafana rules become monitors on a query column.
• What you drop: anything that depends on live metrics, logs or traces.

The practical path is to keep Grafana for infrastructure and add arcodash for the business tables. Most teams that do both don’t need to move anything.

Questions

On price, the difference is the model: arcodash is fixed, Grafana Cloud varies with users and usage. If the real problem is catching a stalled table or a wrong count before someone else spots it, try the free trial of arcodash and point one check at your busiest table.

If you already run Prometheus, the Prometheus connector page covers querying it from arcodash.