Netdata, prometheus and grafana - why recommended in this constellation?

Question

Currently I do setup monitoring-service for our Server. I've found some articles, which recommend to use: Netdata, prometheus and grafana

So I wonder what are the benefits to use all three of them. Can anybody explain it to me, why prometheus and grafana is recommended, if it's seems that netdata do the same stuff as those both?

Even at the netdata-documentations, there are an article about the setup for this: https://learn.netdata.cloud/docs/agent/backends/prometheus/#filtering-metrics-sent-to-prometheus

What are the benefits of this constellation?

Answer 1

In summary, Grafana will display metrics collected by Prometheus from netdata.

There's no benefit to rolling out a stack like that except that Prometheus has scaling capabilities that has made it ubiquitous so if you already have netdata and want to modernize your stack, you can use it together with Prometheus.

If you're starting from scratch, I'd recommend sticking to Grafana + Prometheus + Prometheus' node exporter. This will give you about the same metrics but is "atomic" in that each does a single thing well, and each component is independant so you can administer each one without (exceedingly) impacting the others.

Answer 2

I would suggest, that if you are just starting out, a combination of the Netdata Agent + Netdata Cloud has you covered.

You install the Netdata Agent on every computer that you wish to monitor and you use Netdata Cloud as the unified pane of glass to view the entire infrastructure in a "centralized" way. Note that the data are stored locally, on every node, and they are streamed THROUGH cloud to your browser directly.

If Netdata monitors whatever you need, I don't see the reason to spend time in setting up Prometheus and grafana, as Netdata will give you everything (data-collection, alerts, charts) out of the box.

You can enable exporting in case you want to archive metrics for a long-time (say 6 months+). In that case,you can export every 60s to save up space, while Netdata will keep the per-second metrics for the near term.

Disclaimer: I work at Netdata

Answer 3

You can now drop Prometheus and query/visualize metrics right from the Netdata hose using the new and official Netdata plugin for Grafana.

Answer 4

I've been using Netdata for a couple of years and love it so much. I explain my reasons here:

• Netdata evolves quickly. Compared to other monitoring tools, Netdata is actively evolving and each time I go to their website I see lots of changes. You don't see this for other monitoring tools. Take Prometheus or Zabbix for example. They have very longer release cycles and they rarely add new features.
• Magnificent dashboard. No other monitoring tools give you the dashboard as smooth and beautiful as netdata. It's fast, informative, feature rich. They add information for some chart to teach you what those metrics are and how to deal with them.
• Real-time monitoring. No other monitoring tools gives you real-time monitoring data. Most of them are not designed for real-time monitoring.
• Integrated metrics. Compared to prometheus exporters, netdata have plugins for monitoring different technologies.
• Machine learning feature. Netdata will do anomaly detection based on you resource metric data and give you insight about your infrastructure. You can't find this feature anywhere else this easy and free.
• Log metrics extraction: Netdata reads log files (such as web logs or systemd logs) and extract useful metrics from them which resolves your need for a log management system in most cases!

The only problem with netdata is that it's a full featured monitoring agent installed on a single host. In monitoring we need to collect metrics from different hosts and store them somewhere in a centralized manner in case we need them later or use occasionally. Netdata needs some other tool to send data to and that tool store the historical metrics for longer time so we don't lose data in case our host fails or hard disk becomes faulty. Here comes prometheus (or Netdata Cloud) to solve this need. We can ask prometheus to pull metrics from all our netdata instances and store them centrally. Then we use Grafana to view metrics in beautiful dashboards. I think we can use other tools for storage instead of prometheus if they can support OpenMetric standard.