Yes: Business and up
Signed SLA statements
Each period's attainment, signed and checkable by anyone without an account.
CompareGrafana Cloud
Observer vs Grafana Cloud
Grafana Cloud is where many teams keep their metrics, dashboards, SLOs and on-call. It has no status page of its own; Observer reads the same Prometheus data and gives your customers the page and the record.Checked against Grafana Cloud's own public pages.
Observer ahead on
5of 7
- Signed SLA statementsPartly:
- Per-customer statusNo:
- What drives statusNo:
- Built-in monitoringYes:
- Signal source & data localityPartly:
- Pricing modelPartly:
- Open source & engineer surfaceYes:
Grafana Cloud's answers from its own public pages, checked .
Verdict
The short answer.
Grafana Cloud is built for
Your engineers. Metrics, logs, dashboards, alerting, SLOs and Grafana IRM for on-call, all read inside your team.
Observer instead
Your customers. Observer reads the same Prometheus data with the same PromQL, publishes the status page and SLA statements, and sends status changes to Grafana IRM.
Pick
Keep Grafana Cloud for your engineers. Add Observer when customers need a status page and an SLA record read from the same metrics.
At a glance
The record first, then the mechanics.
What your customers can check comes first. How the page runs follows.
The record
Signed SLA statements
- Observer
- Yes: Business plan
- Grafana Cloud
- Partly: SLO dashboards
Per-customer status
- Observer
- Yes: from Pro
- Grafana Cloud
- No: no status page
The mechanics
What drives status
- Observer
- Yes: metric + threshold
- Grafana Cloud
- No: no status page
Built-in monitoring
- Observer
- Yes: telemetry + probes
- Grafana Cloud
- Yes: full stack
Signal source & data locality
- Observer
- Yes: inside your network
- Grafana Cloud
- Partly: send it your data
Pricing model
- Observer
- Yes: flat per tier
- Grafana Cloud
- Partly: usage + per user
Open source & engineer surface
- Observer
- Yes: agent, Apache 2.0
- Grafana Cloud
- Yes: AGPL-3.0 core
Yes: does itPartly: partlyNo: does notMethod: each answer is graded against a status page with a record customers can check, using the vendor's own public pages. Checked .
Side by side
Line by line.
The same 7 dimensions on every comparison, each with the question it answers.
Dimension
Observer
Grafana Cloud
The record
Signed SLA statements
Can a customer check the SLA figure without taking your word for it?
Yes: Business plan
Each period's attainment, with the working shown, hashed and signed. Your customer checks it at /reliability/verify without an account.
SLO dashboards
Grafana SLO shows error budgets and burn rates to your engineers. Nothing is published for customers to check.
Per-customer status
Can each customer see their own status?
Yes: from Pro
One probe can publish a different verdict per customer, with its own SLO target, subscribers and access.
no status page
No status page, so no per-customer status.
The mechanics
What drives status
What flips a component from up to down?
metric + threshold
Reads the metric your engineers already watch, applies your threshold, and publishes the verdict.
no status page
Alerts and SLOs drive on-call, not a public page. Grafana IRM can open a Statuspage incident when one is declared, but does not set component status.
Built-in monitoring
What does it check on its own?
Yes: telemetry + probes
Prometheus, OpenTelemetry, CloudWatch, Loki, Elasticsearch and SQL/NoSQL, plus HTTP, TCP, DNS, TLS, ICMP, gRPC and WebSocket probes.
Yes: full stack
Metrics, logs, dashboards, alerting and SLOs in Grafana Cloud, with Grafana IRM for on-call.
Signal source & data locality
Where does the signal come from, and where does it go?
inside your network
The agent runs next to your data and sends readings and verdicts outbound. Database and source credentials stay on the host.
send it your data
You send metrics and logs to Grafana Cloud, where its alerts and SLOs read them.
Pricing model
How does the bill grow?
flat per tier
Free, then $19, $79 or $199 a month, then Enterprise. No per-seat or per-subscriber fees.
usage + per user
A free tier, then a monthly platform fee plus usage. Grafana IRM: 3 active users free, then $20 per active user.
Open source & engineer surface
What can engineers read, script and extend?
Yes: agent, Apache 2.0
Open-source agent, public API, CLI, MCP server and AI-assistant skills.
Yes: AGPL-3.0 core
Grafana, Loki and Tempo are AGPL-3.0. Grafana Cloud is hosted by Grafana Labs.
Beyond the page
What Observer adds after the incident.
The reporting customers, procurement and auditors ask for, built on the same measurements as the page.
Yes: Business and up
Reliability Center
Public commitments and attainment, and a private portal per customer.
Yes: Pro and up
Executive dashboard
Attainment, risk and MTTR across every service, for the board.
Yes: Starter and up
SLOs and error budgets
Budgets read from the same metrics, with burn alerts.
Yes: Pro and up
Status per customer
One probe, each customer judged against their own SLO.
Fair to say
When Grafana alone is enough.
Internal SLOs only
Your SLOs and error budgets are for engineers, and no customer asks to see them.
Statuspage already works
You are happy with Statuspage, and Grafana IRM opening an incident there when one is declared covers what you need.
Migration
Adding Observer to Grafana.
Run the agent
One container next to the systems you already monitor. It dials out; nothing in your network accepts a connection from Observer.
Map what you watch
Point the agent at your Grafana Cloud Prometheus endpoint with basic auth. Each panel your team watches becomes a metric: the same PromQL, plus two thresholds.
Publish and route
Group metrics into services and publish the page. Send flips to Slack, PagerDuty, Teams or a webhook; recoveries always deliver.
Switch over
Connect Grafana IRM as an alert channel so status changes reach on-call. Point your status domain at Observer once the page matches what you would have posted (custom domains are on paid plans).
Questions
Asked before switching.
Does Grafana Cloud have a status page?
Not a native one. Grafana IRM integrates with Atlassian Statuspage: it can open a Statuspage incident when an IRM incident is declared and keep the title in sync, but it does not set component status.
Does Observer replace Grafana?
No. Observer sits beside it. The agent reads the Prometheus data you already send to Grafana Cloud, with the same PromQL, and sends status changes to Grafana IRM. Your dashboards, alerts and on-call stay where they are.
How is this different from Grafana SLO?
Grafana SLO gives engineers error budgets, burn rates and alert rules. Observer turns the same metrics into what customers see: the status page, per-customer status and signed SLA statements.
Sources
Where the facts come from.
Every statement about Grafana Cloud on this page was checked against these public pages, most recently on .
Keep reading