[openstack-dev] [TripleO] consistency vs packages in TripleO

Robert Collins robertc at robertcollins.net
Sat Feb 15 03:56:19 UTC 2014


On 15 February 2014 07:46, Clint Byrum <clint at fewbar.com> wrote:

> * Reference implementations are always derided as not realistic. I think
>   we need to think of a different term. I prefer to just say that this
>   is the upstream implementation. We very much expect that a cloud can
>   and should operate with this model unmodified. Packagers are doing so
>   to fit OpenStack into a greater support model, not because "nobody
>   would ever want to run upstream." I think of how a large portion of
>   MySQL users tend to just run upstream's binaries. They don't call this
>   the reference binaries.

Ok, upstream - ack.

> * Documentation can be split into an architecture guide which should be
>   a single source of truth and document interfaces only, and an operations
>   guide, which will focus on the upstream operations story. Distros should
>   provide sub-sections for that story to document their differences.
>   They should not, however, be putting distro specific interfaces in the
>   architecture documentation, and we would reject such things until they
>   are known to work upstream.

Ok.

I'll leave this a few more days to see if more data points arrive, but
it seems largely slanted in this direction.

That said, I wish there were some way to assess the cost/benefits in
terms of OpenStack adoption - which is in a way an operating system
itself - consider VMWare - there's /one/ VMWare, no matter which org
you buy it from, and which addons or integrations or support that org
does.

-Rob

-- 
Robert Collins <rbtcollins at hp.com>
Distinguished Technologist
HP Converged Cloud



More information about the OpenStack-dev mailing list