Back to proposals overview - program

Delightful Autoscaling with Chef and AWS


No single tool does a good job provisioning a complete stack -- from VM container or bare metal, to fully-provisioned server, deployed applications, and multi-node topology.

AWS and Chef are two tools that we often use, and together they can cover the complete stack:

  • AWS handles the infrastructure, up to a running instance.
  • Chef provisions, configures, and deploys on top of the base instance.
  • We use both tools in concert to manage topologies of multiple nodes.

Two common patterns for integrating AWS and Chef are 1) baking Chef into AMIs, and 2) bootstrapping it through CloudFormation and user-data. But both approaches have drawbacks, especially when it comes to autoscaling.
Enter Linecook.

Linecook is a Chef "provisioner" we've created, specifically with autoscaling in mind. It consumes AWS autoscale notifications to rapidly bootstrap newly-created nodes with Chef, and delete terminated ones from Chef Server.

In our presentation, we propose to:

  • Talk about the basic problem (why no single tool covers the entire stack)
  • Explain the limitations of the common patterns for integrating Chef and AWS
  • Demonstrate and discuss the Linecook approach
  • Discuss going beyond AWS: support for other clouds

Speaker: Speaker 20

blog comments powered by Disqus
Compuware Target GovDelivery Dell Software CA Technologies Chef ActiveState SPS Commerce

Silver Lake sponsors

Etsy VictorOps Puppet Labs XebiaLabs Nagios Mirantis

Stone Arch sponsors

TEKsystems Rackspace PagerDuty GitHub Perficient ipHouse

Media sponsors

Velocity Conf O

Host sponsor

University of St. Thomas