On 08/21/2015 06:49 AM, Matt Micene wrote:
Given:
jzb
Given that a great deal of interesting work is going into the Fedora
Atomic host, we'd like to make Atomic the main deliverable/focus for the
Cloud Working Group and Cloud edition.
and
mattdm
For that, we need the Cloud Base to have *internal* visibility and
connections, but it doesn't need to confuse the marketing message.
I think moving the focus of Cloud SIG to focus on Atomic would confuse
folks more than the currently slim messaging around the Cloud SIG
goals. I think CentOS has the right approach with a separate Atomic
SIG, and I propose that we follow that model. Split Atomic into it's
own SIG, continue the Cloud SIG to focus on cloud things.
The question is "what cloud things"? In talking to other folks about
this - e.g. what could or should we be doing to make cloud more
interesting - I haven't heard a lot of ideas.
The message right now is "we have an image you can use, it does the
standard things... we're kind of out of ideas to make it more interesting."
Atomic is a new way of doing everything. OS management changes,
package
management changes, Docker + K8S + Nulecule + Atomicapp + ? + ? + ?. We
are trying to move at a much more rapid pace than the rest of the Fedora
Project products (see the 2 week release proposal). We are breaking
things at a much more rapid pace than other products as well. Lots of
initial answers on ask / IRC wind up as, "update the tree see if it's
still broken".
I don't think that if you say the word "cloud" in a room of IT folks
today, that over 1% are going to think Atomic. They think OpenStack,
OpenShift, CloudFoundry, Eucalyptus, AWS, ownCloud, etc. Fedora as
foundation, Fedora as tenant, Fedora as a Service! The new Cloud SIG
draft reflects those use cases:
https://fedoraproject.org/wiki/Cloud_SIG-new-draft. Fedora saying "well
when *we* say cloud we mean Atomic not what you think" and having to
explain it doesn't sound like a win.
That draft is from 2013. It's a bit out of date.
Note name of sig != name of deliverable. There's no reason we have to
keep a strong association with the name "cloud" on the site / marketing
materials.
But I'd also disagree we need to spin up a new SIG around this when the
mapping of the Cloud SIG and Atomic interest is close to (if not
exactly) 1:1.
Note also there's not a lot of work in things like OpenStack on Fedora
because it's so fast moving. The RDO folks tend to target CentOS rather
than Fedora, though I *think* that running on Fedora is also w/in scope.
(I'm CC'ing Rich Bowen here b/c he's far more in the know about our
OpenStack/RDO efforts than I.)
And while there's overlap in uses like cloud-init, Atomic is much
more
likely to want a completely new compatible implementation because the
dependency tree makes maintenance more complex than a Fedora OpenStack
guest image will need to care about. (See my previous rants on sizing
;-) ). Or a Fedora Server optimized for running nova-compute. The
Cloud Base Image and the Atomic host or the Docker Base Image don't and
shouldn't have much in common because the use cases are different.
I do think that the Cloud SIG does need better messaging about its
intents and goals. The new SIG draft looks like a step in the right
direction to me. Define major focus areas (run IaaS on Fedora, run
Fedora in IaaS, etc) then layer in use cases and projects from the group.
Again - it's several years old, so I'm not sure that the "new" draft is
applicable here. :-)
A new Atomic SIG would focus on stabilizing the Atomic host, the
delivery process, and the Docker base image. Atomic hosts are all about
running Docker workloads, so that makes sense to couple. As
containerization grows, I could see the need for moving Docker image
management somewhere else.
Did I miss the 72 hr window?
Not quite. Other thoughts?
Best,
jzb
--
Joe Brockmeier | Community Team, OSAS
jzb(a)redhat.com |
http://community.redhat.com/
Twitter: @jzb |
http://dissociatedpress.net/