The traditional “Cluster” architecture has been around the web since the 1990s. This tried and true model has supported 1,000s of large sites on any number of platforms. However, it has some serious drawbacks: these patterns lead to a high cost per site, are still fairly inflexible, have relatively low ceiling on scale, and retain some distressingly "rube-goldberg-esque" aspects.

Modern web infrastructure models break out of these traps by favoring more uniform systems for deploying applications. Leaders in this space include Google (who’s internal hosting resource is actually called “borg”) and Heroku (who wanted to call theirs “borg” until they found out that’s the name Google used).

Cloud infrastructure is accelerating this advance by democratizing the scale of computing necessary to use these techniques. If we’re to reach a future where Drupal is powering millions of sites and globally available for anyone to start using with minimal cost/friction, creating a Borg modal for Drupal would be a great help.

Join us as we show you how we’ve applied these concepts at Pantheon.

Intended Audience

This session is intended for people interested in helping Drupal succeed as a global platform, as well as those interested in web-scale architecture generally.

Questions Answered

• What are the differences between a cluster and a borg or grid system? What advantages and disadvantages do they offer.
• How do the particular characteristics of Drupal play into building this kind of architecture?
• What’s inside the black box? How did you guys build this?

Audio available here:
http://www.archive.org/details/InternetScaleBuildingThedrupalBorg

Slides here:
http://2011.badcamp.net/sites/default/files/slides/pantheon_DROPs.pdf

Slides: 

pantheon_DROPs.pdf