5 ms·

NetData: monitor your systems and applications, with interactive web dashboards

115 points by sandebert 8y ago

fgheorghe 8y ago
How is this better than newrelic?
- askz 8y ago
  I think, with my miserable knowledge of NewRelic, that NR is only for application monitoring. Not for system too, as this project.
  - dudleypippin 8y ago
    New Relic absolutely does system monitoring. It's gonna cost you, however...
- eddyg 8y ago
  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."
  - fgheorghe 8y ago
    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.
- a012 8y ago
  Newrelic is very expensive when you have hundreds of hosts (or hundreds of pods of you're running k8s)
- polbatllo 8y ago
  It's free.
  - fgheorghe 8y ago
    "It's free". It costs sysadmins time to configure and maintain. Probably NewRelic costs less, depending on where the salary is paid.
eddyg 8y ago
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. [0]https://news.ycombinator.com/submitted?id=ktsaou https://news.ycombinator.com/submitted?id=ktsaou
danielneri 8y ago
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. NetData looks great, though!
- manigandham 8y ago
  With 30k github stars and all those (automatic) integrations, including the backing databases you mentioned, it doesn't seem that niche.
- nammi 8y ago
  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. https://github.com/firehol/netdata/wiki/netdata-backends 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
radiowave 8y ago
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.
manigandham 8y ago
This is a seriously impressive project, and a great interactive infographic to explain it all: https://my-netdata.io/infographic.html https://my-netdata.io/infographic.html
petecooper 8y ago
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.
- leetbulb 8y ago
  Same here. It's on every standalone server I manage. I also use it on my workstation.
- nammi 8y ago
  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. https://github.com/firehol/netdata/wiki/netdata-backends https://github.com/firehol/netdata/wiki/netdata-backends
  - manigandham 8y ago
    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. https://github.com/firehol/netdata/wiki/Memory-Requirements https://github.com/firehol/netdata/wiki/Memory-Requirements
    - nammi 8y ago
      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
PappaPatat 8y ago
Incredibly beautiful and easy to install on a wide range of linux distros including Raspberry with its standard image. 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!

115 points by sandebert 8y ago

fgheorghe 8y ago

How is this better than newrelic?

askz 8y ago
I think, with my miserable knowledge of NewRelic, that NR is only for application monitoring. Not for system too, as this project.
- dudleypippin 8y ago
  New Relic absolutely does system monitoring. It's gonna cost you, however...
eddyg 8y ago
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."
- fgheorghe 8y ago
  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.
a012 8y ago
Newrelic is very expensive when you have hundreds of hosts (or hundreds of pods of you're running k8s)
polbatllo 8y ago
It's free.
- fgheorghe 8y ago
  "It's free". It costs sysadmins time to configure and maintain. Probably NewRelic costs less, depending on where the salary is paid.

eddyg 8y ago

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. [0]https://news.ycombinator.com/submitted?id=ktsaou https://news.ycombinator.com/submitted?id=ktsaou

danielneri 8y ago

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. NetData looks great, though!

manigandham 8y ago
With 30k github stars and all those (automatic) integrations, including the backing databases you mentioned, it doesn't seem that niche.
nammi 8y ago
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. https://github.com/firehol/netdata/wiki/netdata-backends 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

radiowave 8y ago

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.

manigandham 8y ago

This is a seriously impressive project, and a great interactive infographic to explain it all: https://my-netdata.io/infographic.html https://my-netdata.io/infographic.html

petecooper 8y ago

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.

leetbulb 8y ago
Same here. It's on every standalone server I manage. I also use it on my workstation.
nammi 8y ago
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. https://github.com/firehol/netdata/wiki/netdata-backends https://github.com/firehol/netdata/wiki/netdata-backends
- manigandham 8y ago
  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. https://github.com/firehol/netdata/wiki/Memory-Requirements https://github.com/firehol/netdata/wiki/Memory-Requirements
  - nammi 8y ago
    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

PappaPatat 8y ago

Incredibly beautiful and easy to install on a wide range of linux distros including Raspberry with its standard image. 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!