AWSOME has the potential to be really awesome

Earlier today, Canonical announced AWSOME

I'm honestly pretty stoked about it. At the last UDS, Mark Shuttleworth expressed his concern over the state of AWS compatibility and how important it was.  As part of follow up conversations, Vish, Gustavo and I talked about ways to address it, and the one which everyone was the most pleased with was the idea of a gateway service that could talk AWS on the one side and OS API on the other. This would allow the code paths inside of Nova to become simpler, and innovation at the API layer inside of OpenStack could proceed as architecture dictated. At the same time, as a separate project, developers on a gateway wouldn't have to get nova core devs to care about AWS APIs at all, and themselves could write the service to be as robust and full-featured as they want.

Decoupling FTW!

Anyway - I'm at conferences a lot and I have conversations with people alot about ways in which they can help to solve the problems they are experiencing. It's not nearly so common that people step up to the plate and just fully own doing something about it - so I'm extra excited about AWSOME's existence for that reason alone.

At the moment it's listed on Launchpad as being AGPL (although this is not reflected in the source tree), which obviously would exclude it from being an official part of OpenStack, and probably from being deployed on any of the public clouds being stood up. However, if the details of getting it licensed Apache can get worked out, I would certainly personally support including it in OpenStack.

Thanks for the work Canonical! I'm excited to poke/learn more.


  1. [1]   Anonymous
    April 13, 2012 at 03:20 PM

    Why wouldn't you support including it in OpenStack under the AGPL license?
  2. [2]   Monty Taylor
    April 13, 2012 at 03:40 PM

    OpenStack is all under the Apache license, so AGPL code does not fit the project criteria. Additionally, I know that several of the large contributors to OpenStack have policies against deploying AGPL code. I'm not saying that I'm personally against AGPL - just that it's a non-starter for OpenStack.
  3. [3]   Jef Spaleta
    April 13, 2012 at 04:30 PM

    I would have to assume that the licensing choice was delibrately done by Canonical in order to make it highly unlikely for it to taken into the Openstack project. Canonical knows full well what the licensing burden is for a good faith attempt at OpenStack contribution. If their intention was to contribute this into the upstream project then they'd have licensed it accordingly. This is just vendor a new round of differentiation, similar to what they've done with GNOME in the past.