Open Stack

On 09/09/2016 06:22 AM, John Davidge wrote:
> Thierry Carrez wrote:
>
>> [...]
>> In the last years there were a lot of "questions" asked by random
>> contributors, especially around the "One OpenStack" principle (which
>> seems to fuel most of the reaction here). Remarks like "we should really
>> decide once and for all if OpenStack is a collection of independent
>> projects, or one thing".
>>
>> A lot of people actually ignore that this question was already asked,
>> pretty early on (by John Dickinson in June 2011). Back then it was
>> settled by the PPB (the ancestor to the TC). You can read it all
>> here[1]. It was never brought again as a proposed change to the TC, so
>> that decision from June 2011 is still defining how we should think about
>> OpenStack.
>>
>> Most of the TC members know the governance history and know those
>> principles. That is, after all, one of the reasons you elect them for.
>> But we realized that the people asking those questions again and again
>> were not at fault. It was our failure to *document* this history
>> properly which caused the issue. Took us some time to gather the courage
>> to write it, then finally Monty wrote a draft, and I turned it into a
>> change.
>
> To provide a counter point, I think the reason this question is asked so
> often is not because the TC has failed to *document* this policy, but that
> it has failed to *comply* with it.

The TC doesn't comply with anything at all. It's the body that is 
elected to make overarching governance decisions for the OpenStack 
community.

> I¹m of course referring to the introduction of The Big Tent.

And I'm of course going to correct your misstatements below about what 
the Big Tent (formally called the Project Structure Reform) was about.

 > This is the
> moment that OpenStack stopped being: ³A single product made of a lot of
> independent, but cooperating, components.² And became: ³A collection of
> independent projects that work together for some level of integration and
> releases.²

You are mistaken, I'm sorry to say.

Firstly, your statement that OpenStack stopped being "a single product" 
at some point is just flat-out wrong. It never *was* a single product, 
regardless of whether some time in 2011 seven individuals on the project 
policy board said that that was the direction they preferred the 
community to go in.

OpenStack was never a single product made of a lot of independent but 
cooperating components. From the very beginning of OpenStack-time, Swift 
and Nova were never designed or planned as a single product. To this 
*day*, Swift is its own product with very few pieces of integration with 
some other OpenStack projects where relevant (Keystone) but there was 
never any push or reason to have Swift become simply a "part of a single 
OpenStack product".

Secondly, the Big Tent was about changing the process by which new 
projects applying to become "OpenStack projects" were evaluated by the 
TC. We went from a situation of evaluating new projects using 
subjective, often contradicting and changing opinions on architectural 
and software design to instead evaluating new project teams on whether 
the project team followed "The OpenStack Way" (followed the 4 Opens and 
furthered the mission of OpenStack)

The Big Tent **was not a redefinition of what OpenStack was or is**.

> This is in direct contradiction to the stated and collectively understood
> goal of the project, and has left many scratching their heads.

I don't know what you are referring to above as "the stated and 
collectively understood goal of the project". Are you referring to the 
OpenStack Mission Statement? If so, see point above. The Big Tent didn't 
change the Mission nor did it redefine what OpenStack was.

> The principles as written in the review do not accurately describe the
> current state of the project. To make it so that they do, we either need
> to change the principles or change the project. As I see it, our options
> are:
>
> 1. Adjust the project to reflect the principles by abolishing The Big Tent.
> 2. Adjust the principles to reflect the project by redefining it as: ³A
> collection of independent projects that work together for some level of
> integration and releases.²
>
> Personally, my vote is for option 1.

Sounds to me like you are just complaining about OpenStack having too 
many projects in it. If so, please tell us which projects you would have 
leave OpenStack.

No more deployment projects? Cut Triple-O, Fuel, Puppet, Chef, Ansible, 
Saltstack, devstack.

No more competing implementations of things? Cut Ceilometer or Monasca. 
Your choice.

No more Telco-specific stuff? Cut Tacker, networking-sfc, and a whole 
host of other stuff.

My vote is definitely for something #2-like, as I've said before and on 
the review, I believe OpenStack should be a "cloud toolkit" composed of 
well-scoped and limited services in the vein of the UNIX model of do one 
thing and do it well. I believe that vendors and cloud providers should 
be able to choose services and tools from this OpenStack cloud toolkit 
to build clouds, cloud products, and products that utilize OpenStack 
service APIs to please customers.

It makes sense for many of those cloud toolkit services and components 
to integrate well with each other via public, stable interfaces and 
there's nothing about OpenStack being a collection of cloud tools that 
prevents or discourages that integration.

Best,
-jay