Open Stack

On Mon, 23 Apr 2018, Doug Hellmann wrote:
> Excerpts from Chris Dent's message of 2018-04-23 12:09:42 +0100:
>> I'd like to see us work harder to refine the long term goals we are
>> trying to satisfy with the projects that make up OpenStack. This
>> will require us to continue the never-ending discussion about
>> whether OpenStack is a "Software Defined Infrastructure Framework"
>> or a "Cloud Solution" (plenty of people talk the latter, but plenty
>> of other people are spending energy on the former). And then
>
> Do you consider those two approaches to be mutually exclusive?

No, but I do think how we balance and think about them them helps
us understand how to make progress.

> In the past our community has had trouble defining "infrastructure"
> in a way that satisfies everyone. Some people stop at "allocating
> what you need to run a VM" while others consider it closer to
> "everything you need to run an application".
>
> How do you define "infrastructure"?

In this context I'm thinking of infrastructure in terms of plumbing,
using the plumbing and porcelain metaphor sometimes associted with
git:

     https://git-scm.com/book/en/v2/Git-Internals-Plumbing-and-Porcelain

So in your terms "allocating what you need to run a VM" but with
some tweaks.

In Zane's response on this topic, he talks a lot about the
applications that are using the "cloud" and some of the additional
tooling that is needed to satisfy that (some of which might be
considered porcelain). Since my introduction to OpenStack that's
what I've hoped we are trying to build. An Open Source "Cloud
Solution" that enables a healthy application environment that is a
clear and complete alternative to the big three clouds.

However, over the years it has become increasingly evident that a
great deal of our energy is spent working at a different angle to
enable software defined data centers (including data centers that
are decomposed to the edge) that are hyper-aware of hardware and
networks and making that hardware available in the most cost
effective way possible. That's a useful thing to do but our
attention to it is not well aligned with building elastic web
services.

(In this particular case I'm speaking from experience perhaps
overly informed by Nova, where so much work is devoted to
NFV-related use cases. To such extent that people joke about
hardware defined software.)

While I don't think we need to say that we are doing one thing or
the other, we may make some decisions easier by being more willing
to identify which domain or perspective we are thinking about in any
given decision.

-- 
Chris Dent                       ٩◔̯◔۶           https://anticdent.org/
freenode: cdent                                         tw: @anticdent