netdata
real-time performance monitoring
scaled out!
pick a netdata demo server
these demo servers show what you will get by installing netdata
this page is a custom netdata dashboard
charts are coming from 8 servers, in parallel
the servers are not aware of this multi-server dashboard,
each server is not aware of the other servers,
but on this dashboard they are one!
hover on a chart below, or drag it to show the past - the others will follow!
double click on a chart to reset them all
our nginx performance
(we proxy netdata through nginx, on the demo sites)
EU - London web requests/s
US - Atlanta web requests/s
US - California web requests/s
EU - London active connections
US - Atlanta active connections
US - California active connections
Canada active connections
these charts are draggable and touchable, double click them to reset them
bandwidth consumption on the demo sites
Linux QoS is configured by
FireQOS
these legends are interactive and the charts are resizable here ^^^
DDoS protection performance on the demo sites
iptables SYNPROXY configured by
FireHOL
EU - London, TCP SYN packets/s received
US - Atlanta, TCP SYN packets/s received
US - California, TCP SYN packets/s received
Canada, TCP SYN packets/s received
did you notice the decimal numbers?
netdata interpolates collected values at second boundaries, with nanosecond detail!
CPU Utilization of the demo sites
what is using so much CPU?
The site iplists.firehol.org is maintained by FireHOL - the CPU is used for comparing security IP Lists.
Netdata performance
netdata monitors users, user groups, applications (process trees)
containers (lxc, docker, etc.) and SNMP devices.
EU - London, CPU % of a single core
US - Atlanta, CPU % of a single core
US - California, CPU % of a single core
Toronto, CPU % of a single core
this utilization is about the whole netdata process tree and the percentage is of a single core!
including BASH plugins (it monitors mysql on the demo sites), node.js plugins (it monitors bind9 on the demo sites), etc.
and including the chart refreshes for the dashboards of all viewers.
EU - London, API average response time in milliseconds
US - Atlanta, API average response time in milliseconds
US - California, API average response time in milliseconds
Canada, API average response time in milliseconds
netdata is really fast (the values are milliseconds!)
These values include everything, from the reception of the first byte to the dispatch of the last, including gzip compression.
Values above 2-3ms are usually chart refreshes of charts with several dimensions, charts with very long durations (zoomed out), or file transfers.