Sreekanth Vadagiri
Profile_image?user_id=15177334&size=bigger
Ranked 35 in Phase 1 with 49 unique views, 17 counted upvotes and 7 counted downvotes

Deploying Rails in Large Scale

Simple Rails deployments are easy. Create a couple of machines by hand, add them to Capistrano script and let it do the magic for you. This sort of works if the number of machines you manage is small, but as soon as the complexity and the number of machines increase to deal with scale this approach breaks down.

This talk focuses on dealing with rails deployments for large web apps that spread into 10s or 100s of machines, and how to deal with aspects of scale that is not a big deal with small deployments.

Some of the following

  • Zero down time deployments
  • Managing Redundancies/Backups/slaves (across DC's)
  • Capistrano's issues
  • AutoScaling

Other things I may cover

  • Paas vs IaaS debate(in context of large apps)
  • Puppet vs Chef debate( in context of large apps)

A lot of this talk comes from experiences working on such large web apps. If time allows share some Architecural wins, faux pas, and some war stories.


Previous Next