----- Original Message -----
From: "Eric Helms" <ehelms(a)redhat.com>
To: "Hugh O. Brock" <hbrock(a)redhat.com>
Cc: aeolus-devel(a)lists.fedorahosted.org, "Greg Blomquist"
<gblomqui(a)redhat.com>
Sent: Monday, October 29, 2012 7:35:34 PM
Subject: Re: Winged Monkey
I have one follow-up question that popped up into my head. The
differences you outlined are clear to me from a the perspective of a
user or organization who is setting up a PaaS or IaaS for
consumption by some group of users. From a user perspective, who is
looking at functionality and what a UI provides, the distinction
feels less clear to me. Going from OpenShift to Winged Monkey feels
in some respects like simply being granted more control. In other
words, as a basic user the OpenShift interface provides a very
stream-lined and simplified approach with regards to here is a
pre-configured environment that let's me do straight-forward
applications really easy. Now, as I move from being a basic user to
one that needs more control, or needs a more complex environment I
flip the switch over to a Winged Monkey interface where I now have
more options presented to me, more control and the tools to handle
my more complex setups. However, I can also see it from the reverse
standpoint after re-reading Greg's response, where you would use
Winged Monkey to test and configure an environment, an application
stack to a stable point that you could then bundle and present to
users via a PaaS.
Which finally gets me to my question - is there a project goal to
achieve both or one direction in particular?
Perhaps there's an eventual convergence of Openshift Cartridge
development and the concept of system configuration management and
maintenance.
I don't know of any projects that are specifically attacking this
problem. They might exist, I just haven't heard.
Also, while I think we're quick to adopt new ideas and push things
forward, I'm not certain that consumers are ready for such a
convergence yet. Maybe I'm wrong (I often am:), but I think that
there's still a large segment of the potential market that hasn't
even begun to adopt the basis of cloud computing yet. Let alone,
understanding the subtle differences between something like IaaS
and PaaS. Or, even understanding why they want those differences
to melt away.
I believe there are still several evolutionary steps that need to
take place before the majority of consumers can get to that point.
Not to say there aren't some right now. But, I honestly think those
that are ready for it are the same guys pounding out the code and
are close enough to the problem to see the contrasts and similarities.
Sorry for rambling. But, it's a very interesting point!
-Eric
----- Original Message -----
From: "Hugh Brock" <hbrock(a)redhat.com>
To: "Greg Blomquist" <gblomqui(a)redhat.com>
Cc: "Eric Helms" <ehelms(a)redhat.com>,
aeolus-devel(a)lists.fedorahosted.org
Sent: Monday, October 29, 2012 5:49:35 PM
Subject: Re: Winged Monkey
On Mon, Oct 29, 2012 at 05:37:48PM -0400, Greg Blomquist wrote:
> Thanks for following along, Eric!
>
> ----- Original Message -----
> > From: "Eric Helms" <ehelms(a)redhat.com>
> > To: aeolus-devel(a)lists.fedorahosted.org
> > Sent: Monday, October 29, 2012 3:28:46 PM
> > Subject: Re: Winged Monkey
> >
> >
> > The question: As a user of Openshift, aspects of the design and
> > the
> > feel
> > of the UI feel 1-for-1 and in some respects are attempting to
> > meet
> > the
> > same needs. On which points are there clear divisions between
> > the
> > two
> > projects? Would I be choosing based on maturity and feature sets
> > down
> > the road or is there a clear 'Openshift fills these needs' while
> > 'Winged
> > Monkey fills these other needs'?
> >
>
> This drives to the heart of the distinction between IaaS and PaaS.
>
> In the long term (and maybe for certain user classes), this
> distinction
> may become completely blurred.
>
> However, for the near term (3 years? 5 years? longer?) the
> distinction is
> still visible. With PaaS, a consumer is getting a pre-packaged
> platform
> environment where they can run their specific application (rails,
> django,
> whatever). The tradeoff is the amount of control the consumer will
> have
> with the environment they are allotted.
>
> With IaaS, the consumer requests a set of resources based on some
> specification, and will generally have larger control over those
> resources
> than allowed under PaaS.
>
> So, whereas OpenShift will supply a user with an environment to run
> a
> Django app, Winged Monkey will supply a user with a resource for
> testing
> out installing and configuring Django from scratch.
>
> Hope that helps. Lemme know if that doesn't really clear it up.
Yeah I just want to reinforce what Greg said here.
I think (based on my own recent personal experience with this) that
it
will be a long time before most people are willing to hand over the
keys
to their production apps to the extent you have to do with a PaaS
solution. I'm talking about tight control over software versions,
network configuration, instance sizing, and so on -- all stuff you
take
for granted with IaaS, but that PaaS deliberately obscures from you.
This doesn't mean PaaS isn't useful, far far from it. But I think
there
are always going to be people who want root on all their VMs, and
those
people are going to want a solid management tool that makes managing
them easy across platforms. This is what we're banking on with Winged
Monkey.
Take care,
--Hugh
--
== Hugh Brock, hbrock(a)redhat.com ==
== Engineering Manager, Cloud BU ==
== Aeolus Project: Manage virtual infrastructure across clouds. ==
==
http://aeolusproject.org ==
"I know that you believe you understand what you think I said, but
I’m
not sure you realize that what you heard is not what I meant."
--Robert McCloskey