11. MORE IDEAS (1)
• Index massive amounts of RRD files in
a database for faster lookups
• Working prototype with 200,000
datasources running in production
• Billions of datapoints on a single server
12. MORE IDEAS (2)
• A customer asked for reports like:
• Graph all (1000s) 3G/UMTS links
• "Worst ones first"
• (means order them by antenna gain, using the
daily/weekly/monthly standard deviation)
• RRD is still one of the fastest NoSQL
databases for time series data
15. • Not designed to replace Grafana
• Fills the gap where Grafana fails
• Multiple (many) instances of one thing
• Lots of information at first sight
• Many CPUs, hundreds of interfaces
18. ROADMAP
Allow one to configure distinct trees
(Icinga2 hosts, services, CollectD...)
• Ship nice templates for ITL checks
• Will be released very soon
l
Needs Icinga 2.4
19. WHAT ELSE?
• Implementing a basic hook is easy
• We'd like to see support for more tools
• Think of Munin, Cacti, Prometheus...
27. • Hundreds of nodes are not an issue
• Thousands also, but they hurt your web
browser
• It's therefore perfectly valid to define
sub-nodes referencing processes
defined elsewhere
29. • Alert based on business processes
• Generate availability reports (SLA)
• How? Define Icinga checks!
• Hint: use Puppet to generate processes
31. ROADMAP
• Publicly available since a long time
• TBD: improve the process editor
• Stable release within next weeks
46. ROADMAP
• Implement an Elastic Search query
renderer for our powerful filters
• Will allows us to enforce ACLs
• We want to have it soon, but no
concrete schedule right now
58. Target audience
• Point & click users
• Fully automated environments
• At the very same time (!!)
59. Icinga loves automation
• Perfectly fit for Puppet
• Collect exported resources or query
PDB to build Icinga config
• You basically define hosts equipped
with custom vars
• Services are mostly apply rules based
on host properties
60. Monitoring has to „just work“
• No one wants to waste time on it
• But not every system is fully automated
• e.g. „Add a new MSSQL instance“
• Environmental sensors
61. Deploying every few minutes?
• Don't want to wait for next Puppet run?
• New hosts or apps need to be actively
monitored seconds after being deployed