Opscode Chef Short-Term Roadmap and Performance Improvements

Session Title:

Convener:
Christopher Brown (cb@opscode.com, @skeptomai), Kevin Smith (k@opscode.com, @kevsmith)

Participants:
A cast of thousands... Well, at least a couple dozen and Sean Horn

Slides:
chef_summit_scaling.pdf

Summary of discussions:

Rob Hirschfeld's Notes

Chef <3 Erlang

Scaling

• moving to more Erlang on REST APIs, made a 2x in memory use difference
• Erlang bits have realy steady performance

Database performance

• they do not like CouchDB for their use cases
• moving to mySQL, but will allow users to pick

Noop/Dry Run mode

• asking for suggestions

Search improvements

• tune for most common: single attribute & single mode
• considering new syntax for simple search
• subsets of data

Reporting

What will we do now? What needs to happen next?

Q from Sean Horn: Are the search improvements mutually exclusive with retaining all currently available search syntax? Could there be two different modes?
I think there will always be someone who will miss the full power, even when 90% don't need it.
A: (from @skeptomai) Some of the search improvements will dovetail right in with the current implementation and will share and augment syntax in a backward-compatible way.
We are thinking of a new feature as well, that fits many use cases, but would have a different syntax.