Before I share my takeaway from this MariaDB observability poll, I would like to thank all participants and highlight that these recent polls are very popular, and your participation makes us happy.
That said, we recently asked the MariaDB community the following question:
Which observability tools do you use for MariaDB?
I like polls like this one.
Not because they are perfect. They are not.
Not because they replace real field experience. They do not.
But because sometimes they confirm, very clearly, what many of us already see in practice.
And with 1,733 votes, the answer is loud and clear.
The winner is not really up for debate
The top results are:
- Prometheus: 50% (858 votes)
- Grafana self-hosted: 36% (626 votes)
After that, everything drops very quickly:
- Other: 5% (94 votes)
- Nagios / Icinga / Checkmk: 3% (50 votes)
- Zabbix: 1% (14 votes)
- Elastic / ELK / OpenSearch: 1% (10 votes)
- OpenTelemetry-based stack: 1% (9 votes)
- Custom in-house solution: 1% (9 votes)
And then the rest is almost anecdotal.
So let’s not overcomplicate it:
For MariaDB observability, Prometheus and Grafana have become the default choice.
Honestly, this is exactly what I expected.
This is what the community actually trusts
There is always a lot of noise in observability.
Every year, there is a new trend, a new platform, a new promise of “single pane of glass,” a new AI-powered dashboard that is supposed to tell you everything before you even ask.
But when people have to monitor real MariaDB servers, in real environments, under real pressure, they seem to go back to what works:
- metrics
- dashboards
- control
- simplicity
- flexibility
And that is where Prometheus fits extremely well.
It is not fancy.
It is not trying to be magical.
It just does the job.
And for databases, that matters. Aren’t DBAs boring anyway? 😉
The self-hosted Grafana result is probably my favorite one
One number really stood out to me:
- Grafana self-hosted: 626 votes
- Grafana Cloud: 5 votes
That is not a small difference. That is a statement.
It tells me that MariaDB users still want to keep observability close to the database. They want to run it, manage it, and understand it themselves.
I am not surprised; this is my favorite approach, and I love Grafana.
MariaDB is very often used by people and companies who care about infrastructure choices. They care about cost. They care about control. They care about data locality. They care about not being forced into some opaque SaaS model for everything.
And database monitoring is one of those areas where self-hosting still makes a lot of sense for our users.
They want to keep their metrics, build their dashboards, control their alerts, and not necessarily want to depend on someone else’s pricing model to get them. That feels very aligned with the MariaDB ecosystem. At least to the ecosystem answering our polls!
Traditional monitoring is still there, but it is no longer driving the conversation
Seeing Nagios / Icinga / Checkmk and Zabbix still present is not shocking at all.
Those tools are part of the history of infrastructure monitoring. Many teams still have them. Some teams rely on them heavily. In some companies, they are deeply embedded in operational processes, and nobody is going to rip them out tomorrow.
But the poll also shows something important: they are no longer the center.
That world was built around checks, thresholds, and service states.
The modern expectation for MariaDB observability is different.
Now people want trends.
They want time-series data.
They want correlation.
They want dashboards that let them explore, not just alerts that tell them something is already wrong.
That is a different operational mindset. And as a DBA, I always say that trending graphs are always much more significant than a threshold alert.
The commercial platforms barely register
Datadog, New Relic, SolarWinds, cloud vendor monitoring only… they are almost invisible in this poll.
That does not mean they are absent from the market. Of course, they are not.
But in this audience, for this use case, they are clearly not what comes first.
And I think there is a reason for that.
When people monitor MariaDB seriously, they often want direct visibility into what the server is doing. They want proper metrics. They want to build their own dashboards. They want to understand the internals. They do not want a polished abstraction layer that hides the interesting part.
DBAs and database engineers usually don’t want fewer details.
They are looking for the right details.
OpenTelemetry is interesting, but clearly not there yet
Only 1% selected an OpenTelemetry-based stack.
This is worth noticing because OpenTelemetry is everywhere in observability discussions right now. It is the modern answer to many questions, especially on the application side.
But for MariaDB monitoring, the poll suggests something very simple:
not yet.
At least not as a dominant operational choice.
That makes sense to me. MariaDB monitoring is still very metric-centric. The Prometheus ecosystem is mature, practical, and already deeply integrated into many environments. Exporters, alerting, dashboards, recording rules — all that already exists and is well understood.
Maybe OpenTelemetry will become more important later.
But today, for MariaDB, the community seems to be saying: “Thanks, but Prometheus already solves the problem.”
I also think this is because MariaDB Server doesn’t yet provide a wide range of observability metrics to collect and follow the telemetry workflow expected by OpenTelemetry. Maybe this is something to improve, could be a nice contribution!
Polls are imperfect, but this one is still very telling
Of course, a poll like this has limits.
Some voters probably selected Prometheus even though they also use Grafana, Alertmanager, and something else. Others may use multiple stacks in parallel. The results do not describe architecture in detail.
But that is not really the point.
The point is to understand what the community sees as its primary observability model.
And on that, I think the message is crystal clear.
My conclusion
If you ask me what this poll says, my answer is simple:
The MariaDB community has already made its choice.
The community is not chasing every new observability fashion. It is not getting distracted by shiny platforms or overcomplicated stories.
It is choosing tools that help people operate databases reliably.
And in the end, that is the only thing that matters.
So yes, the result is very clear:
For MariaDB observability, Prometheus is the standard.
Grafana is the natural companion.
And self-hosted still wins by a mile.
Everything else is, at least for now, a distant second.
And honestly?
That sounds exactly right.
If you are interested in some other cool solutions for monitoring, please take a look at our MariaDB Ecosystem Hub!