Sprint Feedback

[Tom Haddon]

On Tue, 2011-06-07 at 17:14 +0100, Mark Shuttleworth wrote:
> On 07/06/11 14:31, Tom Haddon wrote:
> > Sure, we're all abstracting things out and trusting other people with
> > certain parts of the system. I'm not writing all the code that runs on
> > my system, I'm using packaged code, etc., but that doesn't mean you
> > still don't have visibility on the changes being made or having control
> > over what changes you're comfortable with. For instance, managing /etc
> > in bzr allows you to see any configuration changes made by either a
> > package install or a changed kernel setting, or whatever.
> 
> I think the hardest case is partially-managed environments, and Puppet
> is essentially that. Puppet is NOT in total control of the system, you
> are, puppet is helping you exercise that control.
> 
> Part of the leap to intelligent workloads is that one will stop feeling
> like you NEED to see exactly what's in the config. The ensemble state
> (stored in ZooKeeper) is all you need to map. Of course, we're a long
> way from that, and there are probably going to be places where we want
> to use Ensemble in more controlled environments (like the current
> discussions about Orchestra using Ensemble). But in future, I really
> don't think you're going to want to see the config files, unless you are
> debugging the formula. You're going to want to be able to dump and
> recreate state at the Ensemble level, not much lower than that.

So this is interesting, and makes me think of something I'd been
wondering about in regards to Ensemble at UDS, and brought up briefly
with Gustavo then. In some environments you may care mostly about
bringing services up quickly and easily, but at some stage you may need
to fine tune things to get the best performance possible out of the
available resources. If you're not hands-on with the configuration
files, how would you be able to do that? Or is that simply a formula
issue (i.e. the formula knows enough about the "hardware" it's running
on to set things up for optimum performance)? What happens if the
workload changes and you have to alter some settings to accommodate, for
instance? Or would that just fall outside of the purview of Ensemble and
be something you'd need to either configure manually or have a
configuration management tool for that?