Open Stack

There's been some great output coming from the App Developer WG,
especially the cloud benchmark report[1] that sparked a lot of
conversation and even an excellent blog post from Stefano[2].  It was
that blog post in fact that finally served as the wakeup call for me
to actually get involved with this WG (apologies for coming late to
the party!)

In our meeting yesterday I had some questions about who we are
targeting with the "first OpenStack application" effort[3], and on the
call there was agreement that this is still unclear.  I spent some
time looking through the mailing list archives as well but did not
find anything definitive (though PLEASE point me at any conversations
I've missed on this topic).

I'm asking because I've seen a lot of ambiguity around this topic, and
we are not setting ourselves up for success if we are trying to
satisfy "any developer who wants to build something on top of an
OpenStack cloud".  From the perspective of documentation I think what
the API WG is doing is exactly right; collecting all the API
information so the community of OpenStack developers can work towards
a consistent and unified approach.  This also obviously serves as the
reference for developers as they start building an app on an OpenStack
cloud.

I'm hoping on this thread we can get clarity around the particular app
and app developer we hope will benefit from this work.  Mainly that's
because OpenStack is at it's heart an IaaS, not a PaaS.  We have some
tooling that enables cloud app development (like Heat).  Beyond that
though, we get into the weeds of different projects hoping to provide
PaaS bits on top of OpenStack (Murano, Solum and Magnum jump to my
mind).  Each of those provide an abstraction between the application
and some of the API's available on an OpenStack cloud.  So for a
smaller scale application, I think those (or CloudFoundry or
OpenShift) look pretty attractive.  Which is to say - if an app dev
would be better suited using one of those solutions, we should be
clear about when that makes sense vs. building on top of the native
tooling.

My take is that we are targeting very large scale apps (at the far
extreme, think Netflix) vs. your typical LAMP stack (a low-volume
Wordpress blog for instance).  If that is the scale we're talking
about, agreeing on that will better inform the guidance we provide to
the developers.

Alternatively, if we're thinking of smaller apps expected to be
available in a marketplace (like https://apps.openstack.org), that
will obviously impact our message and guidance.

Hopefully we can gather some thoughts and opinions around this on the
mailing list and cover it in a future meeting!

[1] https://docs.google.com/presentation/d/1owWrAtP8vkgTYcd9Tvgyhcv-YHBP2Fs-rr0FV1T-khY/edit#slide=id.gf780219d1_0_121
[2] http://maffulli.net/2016/01/12/developing-apps-on-openstack-is-still-too-complicated/
[3] http://developer.openstack.org/firstapp-libcloud/

-Christopher