Monitoring performance of your OpenStack environment Matthias Runge - PowerPoint PPT Presentation

Monitoring performance of your OpenStack environment Matthias Runge Senior Software Engineer Nov 07, 2017

MUNIN AND PROMETHEUS T wo common monitoring implementations Diferent age (2002 vs 2012) ● Pulling mechanism ● Single server ● Interface to rrdtool vs TS database ● Many more Ganglia ● Cacti ● pcp ●

DATA COLLECTION

A FEW WORDS ABOUT STATSD De facto metering standard >> 10 server implementations ● originated at Flickr, then continued at Etsy ● Metric store: <metric name>:<value>| ● <type> Gauges, meters, timers ○

DATA COLLECTION WITH COLLECTD takes measurements in regular intervals ● ~ 80 plugins ● cpu, memory, network interface collectd plugins for sending data reading data from another instance modular design ● initial release 2005 ● small, but active community around it ● Data store Visualization

DATA STORE

CARBON, WHISPER, (AND GRAPHITE) alternative to using InfuxDB ● whisper: a database library for storing time- ● series data (similar in design to RRD) graphite is for visualization ● carbon is the daemon to listen for TS data ● The image is taken from graphite website

GNOCCHI Statsd protocol ● Storage-Backends ● Index-Backends ● Authentication & Authorization Keystone vs ● basic auth Data aggregation ● openstack metric status ●

VISUALIZATION

GRAFANA graphite/whisper: support built in ● gnocchi: requires a plugin, available from ● grafana

WHAT TO MONITOR Usual suspects such as free memory, IO waits ● Database writes ● Httpd connections, listeners ● Rabbitmq queue length ● Gnocchi status ● Collectd exec plugin ●

CONCLUSION An alternative approach to standard setup collectd for data collection ● gnocchi for time-series database ● grafana for visualization ●

THANK YOU plus google com/ facebook com/redhati +RedHat nc linkedin com/company/red- twitter com/RedHatNe hat ws youtube com/user/RedHatVide os

REFERENCES https://en wikipedia org/wiki/Comparison_of_network_monitoring_systems ● http://gnocchi xyz ● https://github com/b/statsd_spec ● http://code fickr net/2008/10/27/counting-timing/ ●