A Zabbix Believers Story

Jayesh Thakrar Chief Architect, Mikoomi making enterprise monitoring virtual

Mikoomi, 2010

Slide 1

Topics
1. Introduction 2. Comparison : Nagios v/s Zabbix 3. Zabbix : Architecture Overview 4. Zabbix : Browser based GUI 5. Mikoomi : Open-source Value-Add Agents & Consulting Services

Mikoomi, 2010

Slide 2

Introduction

Mikoomi, 2010

Slide 3

How It All Began..

Needed to monitor IT systems - 24x7

Are applications, web servers, databases and other services up? Visibility into current and historical performance and load Quantifying, charting and trending of load, performance and utilization

Needed insight into performance

Tool for HelpDesk (Level-1 Support)

Mikoomi, 2010

Slide 4

Choices: Commercial Players

Mikoomi, 2010

Slide 5

Choices: Nagios & Derivaties

www.groundworkopensource.com

www.shinken-monitoring.org

Mikoomi, 2010

Slide 6

Choices: Other Open Source

http://www.slac.stanford.edu/xorg/nmtf/nmtf-tools.html#contents
Mikoomi, 2010

Slide 7

Top Contenders: Nagios & Zabbix

Mikoomi, 2010

Slide 8

Nagios: Brief Overview

Pros

Popular and well-known Basis for many other open source systems Template-based and object oriented inheritance Based out of Minneapolis, US Boost (?) by RedHat announcement
http://www.nagios.org/news/77-news-announcements/230-nagios-is-redhats-standardalerting-system

Mikoomi, 2010

Slide 9

Nagios: Brief Overview

Cons

Requires significant effort for setup Setup, admin and configuration = text file based Monitoring data stored in single flat file (or via pipe into database) High I/O on data file from monitoring and UI Configuration change require reload Primitive graphing and monitoring UI
Mikoomi, 2010

Slide 10

Zabbix : Brief Overview

Pros

Agent and agent-less monitoring SNMP support Template based Scalable, distributed architecture Built-in UNIX, log-file, SNMP and URL monitoring Easy to extend with plug-ins or agents Active development Database based monitoring data storage Thresholds and alerting separate from monitoring
Mikoomi, 2010

Slide 11

Zabbix : Brief Overview

Pros

Multiple items or attributes per monitored entity Different items of an entity can be monitored by different mechanisms Can define alerts based on comparison of current item value with historical values, averages, etc. Can build dependencies between monitored entities Pre-canned (template-based) graphs as well as adhoc graphs on any monitored item User-defined maps, screens and slide-shows
Mikoomi, 2010

Slide 12

Nagios to Zabbix

N
Mikoomi, 2010

Z
Slide 13

Convinced that N to Z is more than Just a 90 rotation ??

Zabbix Architecture Overview

Mikoomi, 2010

Slide 14

Zabbix Distributed Architecture

External monitoring data collectors
Zabbix Node (Central) Web Server Zabbix Server Zabbix OS Agents

Zabbix Distributed Nodes

Zabbix Database

External Scripts Proxy Servers or Proxy Agents

Mikoomi, 2010

Slide 15

Inside the Zabbix Server

Zabbix Server Processes
watchdog Poller Processes housekeeper Poller Processes alerter Poller Processes poller Poller Processes httppoller Poller Processes discoverer Poller Processes pinger Poller Processes db_config_syncer Poller Processes db_data_syncer Poller Processes nodewatcher Poller Processes timer Poller Processes Pollerescalator Processes

Mikoomi, 2010

Slide 16

Zabbix OS Agent
OS-level agents for most popular platforms

Linux AIX, HP-UX, Solaris MacOS Windows

OS agents can run external programs to complement / enhance monitoring

Mikoomi, 2010

Slide 17

Zabbix Monitoring Approach

Templates

Define new or modify existing templates Contains monitoring data elements called items Contains thresholds (triggers) and actions on item Collection of pre-defined graphs using items
Hosts = monitored entity e.g. hosts, applications, databases, etc. Define new hosts and link to template Customize triggers and actions if necessary

Hosts

Data Collection by Server, Agent or Proxy

Mikoomi, 2010

Slide 18

Zabbix: Built-in Templates

Mikoomi, 2010

Slide 19

Zabbix: Template Items

Mikoomi, 2010

Slide 20

Zabbix: Item Configuration

Mikoomi, 2010

Slide 21

Zabbix Browser based GUI

Mikoomi, 2010

Slide 22

GUI: Login Page

Mikoomi, 2010

Slide 23

GUI: Dashboard

Mikoomi, 2010

Slide 24

GUI: Dashboard Favorites

Mikoomi, 2010

Slide 25

GUI: Dashboard Minimized

Mikoomi, 2010

Slide 26

GUI: Menu Options

Mikoomi, 2010

Slide 27

GUI: Monitoring Data Display - Tabular

Mikoomi, 2010

Slide 28

GUI: Monitoring Data Display - Tabular

Mikoomi, 2010

Slide 29

GUI: Monitoring Data Graphs - Adhoc

Mikoomi, 2010

Slide 30

GUI: Data Graphs Pre-canned

Mikoomi, 2010

Slide 31

GUI: Data Graphs Custom

Mikoomi, 2010

Slide 32

GUI: Templates and Triggers

Mikoomi, 2010

Slide 33

GUI: Trigger Definitions

Mikoomi, 2010

Slide 34

GUI: Alert Listing

Mikoomi, 2010

Slide 35

GUI: Alert Emails

Mikoomi, 2010

Slide 36

GUI: User & Group Administration

Mikoomi, 2010

Slide 37

GUI: Group Security

Mikoomi, 2010

Slide 38

enterprise monitoring made virtual

Mikoomi, 2010

Slide 39

About mikoomi
Mikoomi, the company

Develops, distributes and supports open-source monitoring solutions Provides custom development and consulting around monitoring and high availability Strong believer in open-source as a consumer and as a producer
Mikoomi, 2010

Slide 40

mikoomi Products & Services

Mikoomi Monitoring Agents

Services & Support

Mikoomi value-add

Zabbix Monitoring Framework

Mikoomi, 2010

Slide 41

mikoomi Products - Appliance

Mikoomi Monitoring Appliance

Appliance = virtual machine template Contains Zabbix + Ubuntu + best practices Zabbix = Best open source monitoring Ubuntu = One of the best Linux variants Quick, easy & flexible to deploy Up and running in less than 60 minutes

Mikoomi, 2010

Slide 42

mikoomi Products Agents

Mikoomi Monitoring Agents

Add-on monitoring capabilities for databases, application servers, software components, custom apps Embed deep product-specific expertise and monitoring best practices Covers key health and performance data Open-source makes them extensible Minimally intrusive on monitored entity Java JVM and DB2 released WebSphere, Tomcat, SQL Server, Oracle, ActiveMQ and others planned for release
Mikoomi, 2010

Slide 43

mikoomi Services
Services

Deployment, implementation and training Consulting & custom development Develop custom monitoring for software vendors to help operations and monitoring of their products

Mikoomi, 2010

Slide 44

mikoomi: Sizing and Capacity

Single node (appliance) with 2 CPUs + 2 GB memory supports monitoring a sizable IT environment

10 20 servers + 20 40 databases or instances + 20 40 application instances

Scales horizontally and vertically

Mikoomi, 2010

Slide 45