I've never deployed newrelic, but one of the nice things about netdata is the lack of dependencies and configuration knobs. From the web page:
"It collects more than 5000 metrics automatically, with zero configuration, it has zero dependencies, requires zero maintenance and comes with more than 100 alarms pre-configured to detect common failures, performance and availability issues."
With new relic you issue a single command to install an agent, or an extension, and configure your license key. Takes a few minutes and you don't need to spend expensive devops time configuring yet another "free" tool.
Wow. This has been submitted to HN many times previously (especially by Costa Tsaousis[0], founder of firehol/my-netdata, as "Show HN") but this is the first time it has received more than a handful of upvotes. It is a very nice real-time performance and health monitoring system.
This is neat, but niche monitoring tools often lose out when trying to capture multiple infrastructure/application/business metrics and chart them in a single place. Toolsets like Grafana & a backing DB (Influx, Graphite, etc.) or a service like DataDog or New Relic tend to win out for larger deployments.
I posted this above, but you can get netdata metrics into graphite,opentsdb,prometeheus,etc formats for those databases with minimal work. In that setup netdata is basically a collection agent reporting to whatever database Grafana is reading from.
The cost of infrastructure monitoring is an instant blocker for my work, but if you are a large profitable company with a huge deployment, then yeah netdata probably isn't the right tool for the job
I've used this a little bit - particularly for lab work it's very handy to have a monitoring system where everything just works and there's a web interface available right from the box, without needing to hook it up to any larger system.
I discovered netdata in 2016 and it's become a standard part of my server builds. Fast, low memory, and crucially it's pretty enough to satisfy/inform non-tech people.
I agree that the value it provides is really high for how simple it is to set up.
At first, my only complaint was the relatively short data retention period, but I found that it was really easy to send the data into InfluxDB or Prometheus by following their wiki. That's not netdata's use-case anyway, and it's trivial to set up a backend db when I want to have long term metrics.
The netdata metrics store is in-memory and works well for several hours, but there's a memory-mapped mode to keep larger amounts of data. There's also a future roadmap to eventually add on-disk compressed archives for even longer retention.
oh cool, thanks for sharing that link. In case I wasn't clear, I'm happy with using netdata for short-term data, and hooking up a db in the rare case I want long-term. But that link has a ton of interesting details, can't wait to read later
20 comments
[ 0.19 ms ] story [ 42.5 ms ] thread"It collects more than 5000 metrics automatically, with zero configuration, it has zero dependencies, requires zero maintenance and comes with more than 100 alarms pre-configured to detect common failures, performance and availability issues."
[0]https://news.ycombinator.com/submitted?id=ktsaou
NetData looks great, though!
https://github.com/firehol/netdata/wiki/netdata-backends
The cost of infrastructure monitoring is an instant blocker for my work, but if you are a large profitable company with a huge deployment, then yeah netdata probably isn't the right tool for the job
At first, my only complaint was the relatively short data retention period, but I found that it was really easy to send the data into InfluxDB or Prometheus by following their wiki. That's not netdata's use-case anyway, and it's trivial to set up a backend db when I want to have long term metrics.
https://github.com/firehol/netdata/wiki/netdata-backends
https://github.com/firehol/netdata/wiki/Memory-Requirements
Amazingly there is only 1 (one) host found in Shodan with netdata running on the default port (19999) but plenty on other ports.
Very cool project. I will certainly take a long look at this, thanks for posting!