On 11/14/2012 03:08 PM, Mo Morsi wrote:
On 11/14/2012 11:00 AM, Scott Seago wrote:
>> I would argue the opposite. If its clear we can do 1 instance well, > 1
>> is easily seen as a matter of parallelizing it after that. Most popular
>> projects take the philosophy of doing one simple thing, but doing that
>> thing really well.
> This isn't clear in the "multi-instance systems across clouds" mission
> It's easy to see "if we can do 1 instance we may be able to do >1 in
> the future", sure -- but it does nothing to communicate that "we can
> do >1 _now_"
Perhaps a better way to put it is, if it's clear that if we can do 1
instance well, >1 is parallelizing that "by the end user".
A couple weeks back I showed off a video [1] how to deploy a koji
installation to multiple clouds in 3 minutes, less than the amount of
time to install aeolus-all and run aeolus-configure.
All I did was combine deltacloud with our tdl format to setup instances
w/ the specified config on multiple clouds. If we don't make things that
easy, someone else will, and users will use that and continue to ignore
our project.
[1]
http://www.youtube.com/watch?v=qF2ctg7ItNc&feature=youtu.be
> We took the approach a while back to focus more on multi-instance
> deployments as one of the differentiating factors here. Lots of people
> handle single instances.
Why aren't they as pervasive as they can be? There is no single defacto
open source standard for simply directly deploying to a single generic
cloud infrastructure provider from the command line yet.
We have the capability, the man (and woman!) power, and are in the
perfect position to become that solution. Once that is in place and
widely used, there are a million things that can be done ontop of it.
But we have to take it one step at a time and the current monolithic
conductor-based approach isn't working. So I'm suggesting we cut our
losses, trim the project, and try some different things.
Ahh, but that's a
different discussion entirely. The 'mission statement'
discussion is an attempt to summarize _what we're doing now (and
intending to do)_ i.e. lets get our current mission written down for
others to understand.
It appears that you're suggesting something at a different level --
actually changing our mission (regardless of whether it's written down)
-- i.e. the suggestion that we stop focusing on multi-instance
deployments and go back to supporting single instances better. That's a
discussion that will probably get lost in the details of an existing
discussion whose goal is to "write down what we're all about" -- so it
probably warrants a separate discussion.
That said, we'd need really compelling reason to redefine our mission to
exclude something that we currently support, since it would essentially
require Red Hat to, in future product releases, either abandon support
of Deployments, or take on that bit of functionality without the
community. I agree with the need to separate out the deployments bit
from the single-instance bit -- to allow users to use one or the other,
or both -- but I don't see any obvious reasons to abandon it entirely.
Scott