Asynchronous Real-time Monitoring with MCollective

Jeff McCune (Puppet Labs)
Operations Mission City
Average rating: ***..
(3.20, 10 ratings)

MCollective is a management framework that provides asynchronous remote procedure calls across collections of systems. MCollective is capable of managing systems, collecting performance and inventory data, and integrating with configuration management tools and monitoring frameworks. It is an ideal tool for large scale and simple orchestration and deployment of cloud systems.

Developers and System Administrators are empowered to address systems using meta-data about the system instead of requiring knowledge of each system’s hostname. Indeed the architecture of MCollective relies on systems being addressed by their meta-data, collected from tools such as Facter and Ohai. This makes it an ideal choice for managing systems in the cloud where hostnames can be unpredictable and transient.

In addition to providing an orchestration and deployment solution, MCollective is designed with real-time, asynchronous monitoring of service availability in mind. Traditional monitoring solutions, such as Nagios, use synchronous polling systems to check service state. As services are moved into the cloud, developers can also bring systems up and down with ever increasing velocity. A cloud monitoring system needs to be frequently reconfigured to monitor these new systems. Traditional monitoring systems are either incapable of this level of monitoring scalability or lack the agility needed to rapidly change configuration. Using an asynchronous message bus and publish/subscribe technologies, MCollective provides the ability to automatically start and stop monitoring of a system as it comes up and down in the cloud.

This session will demonstrate the development and use of small Ruby agents to obtain up to date inventory, status and performance information from a collection of systems. Attendees will:

  • Get an introduction to MCollective
  • Understand why asynchronous communication is important when monitoring
  • Discover the benefits of addressing machines by meta-data
  • Learn how to write a monitoring agent for MCollective in Ruby
  • Understand how to remediate performance issues with MCollective and Puppet
Photo of Jeff McCune

Jeff McCune

Puppet Labs

Jeff McCune is a veteran system administrator and current employee of Puppet Labs. He’s been managing Unix systems full time since 2002 and before joining Puppet Labs was responsible for a hosted SaaS datacenter environment. As a member of the Puppet community, he contributed much of the Mac OS X support in Puppet before joining the company full time. Currently, much of his time is spent working with large customers of Puppet Labs designing, deploying, and improving the performance of their cloud and automation infrastructure.

  • Keynote Systems
  • Cisco
  • Google
  • Neustar
  • Betfair
  • Cotendo
  • Rackspace Hosting
  • Akamai
  • Apica
  • dynaTrace
  • Equinix
  • Facebook
  • New Relic
  • Opscode
  • Salesforce.com
  • Yahoo! Inc.
  • AppDynamics
  • Aptimize
  • Blaze
  • CDNetworks
  • Cedexis
  • Citrix Systems
  • Compuware Corporation
  • Dyn Inc.
  • F5 Networks
  • Heroku
  • Percona
  • Quest Software
  • Schooner Information Technology
  • SiteSpect
  • Splunk
  • Strangeloop
  • WatchMouse
  • Zeus Technology
  • Neustar

Sponsorship Opportunities

For information on exhibition and sponsorship opportunities at the conference, contact Yvonne Romaine at yromaine@oreilly.com

Download the Velocity Sponsor/Exhibitor Prospectus

Contact Us

View a complete list of Velocity contacts