Monitoring for the next decade Monitoring 2.0? 2011 - Olivier Jan
About Me Co-founder of CFSL http://www.monitoring-fr.org Writer of the french book «Nagios 3 au cœur de la supervision Open Source» Open Source monitoring solutions architect In charge of R&D @ Capensis Poor english accent!
Contents A road trip experience between Predictions about what you may need in the next decade to be the king of monitoring The nowadays trends, good ideas and tools in Open Source monitoring free software inside.
#1 Big Distributed Architecture Because 1 mn intervall polling will be the standard You have the project to monitor the whole Internet You finally realize that NDO doesn t scale well
Big Setup Mod_Gearman Nagios Daemon Gearman Server Gearman Worker + LiveStatus + Multisite
Big Setup IcingaMQ Based on messages bus (ZeroMQ) Don t need anymore DNX or Mod_Gearman
Big Setup Shinken + LiveStatus + Shinken UI
#2 End User Experience Because You want to know quickly if that service delivered to user just works You cannot put a monitoring agent on Google Mail You know that the sum of the parts monitored is not equal to the whole
End User Experience You know that the sum of the parts monitored is not equal to the whole DNS OK HTTP OK OS OK NETWORK OK APPLICATION/USER OK
End User Experience Cucumber + Watir Behaviour Driven Monitoring Checks functionnalities Completes the technical checks
End User Experience Cucumber + Watir Nagios Collector Cucumber Watir Webdriver CUCUMBER CRITICAL - Critical: 1, Warning: 0, 1 okay passed=1; failed=1; nosteps=0; total=2; time=0
Cucumber Scenarios Abstraction layer between application managers and technicians Based on the expected behaviour You write in your native language Fonctionnalité: Achat appareil photo Scénario: Page d'accueil Etant donné que je suis sur la page d'accueil internationale Quand je clique sur le lien "France" Alors je devrais voir "En ce moment sur grosvendeur.com" Scénario: Accès à la page photos Etant donné que je suis sur la page d'accueil française de grosvendeur Quand je clique sur le lien "PHOTO" Alors je devrais voir "Le Top Ventes"
#3 Aggregation Correlation Because You don t want to fill mailboxes with notifications You want alignement between technical and business monitoring You want to find root cause problem and have impact analysis
Aggregation The tools Correlation Nagios Business Process Addon Multisite and aggregation Shinken and its business views Node Brain
#4 Better Charts Because rrdgraph generated charts are not so easy to read You want real values You want to choose the period you render
Better Graphs HighCharts What do you prefer?
#5 Dashboards Maps Because You want to mix events, performance, availability datas You want top views that lead to detailed views You want big screens in your datacenter
Dashboards Maps The tools Nagvis
Dashboards Maps The tools Shinken
Dashboards Maps
Mixed View
#6 Deliver Reports in minutes, not days Because Everyone needs SLA reports It s time consuming to prepare monitoring reports You want quickly choose the period and the scope on which you have to report It s not what you prefer
Reports Icinga/Jasper Reports integration
#7 Real Log Management Because Polling at a 5mn intervall with a check_logs will be a bit old school You catch too much boring events & miss interesting ones You know the Truth is in the logs
Real Log Logstash Logs transport Logs sanitizing
Real Log Graylog2 Logs centralisation and visual mining
#8 Hypervision Because You certainly answered yes to one of the previous predictions we have seen It completes your existing monitoring solutions and sit on top of them You want to enlarge your monitoring
Foundations Canopsis Events @ heart Sorry no checks Based on AMQP (Active Message Queue Protocol) standard with RabbitMQ Connects to your existing monitoring solutions Renders dashboards and reports you can build from libraries Engines to agregate, correlate differents types of monitoring datas (business rules, inventory )
Canopsis Architecture more on http://www.canopsis.org
Without them CFSL Romuald Fronteau David Guénault Capensis Netways Jean Gabès
Questions What Else? Thanks for your attention