www.nuug.no/.../20100810-cfengine3.ogv

So i promised i would speak english, in case, we don't know who's out there on the far reaches of the

galaxy.

so if that's ok with everyone, i'll continue in english.

so the subject of configuration management

is now pretty well known in the open source community and cfengine as a project has been around

since 1993 as petter said, and it's still going strong and we're not running out of ideas any time soon.

and so it's great to come here and actually summarize what's been going on in the last couple of years.

because the last couple of years of cfengine have probably been one of the most active develeopment periods

of cfengine as a whole.

as some of you probably know, a couple of years ago we started a company to develop and support the future

of cfengine,

and since then it's been more resources to carry the project forward.

so these are exciting times in terms of

development of configuration management.

So i thought i'd just start out by sketching out the problem

of configuration managament for anyone who might not be completely familiar with it, and then i'll go

into some of the history and explain some of the features of cfengine.

so the challenge that i realized i back in 1993, is that we needed to separate the design of systems

from the implementations of the systems.

in pretty much the same way that today we separate design and implementation in many other areas in software

design,

implementation in web design, we have style sheets that take care of the look and feel,

and then we have the implementation in terms of html.

we can think of cfengine and configuration management tools in some ways being like style sheets for o

perating systems somewhat, deciding how they should look.

And the reason to separate them is pretty obvious, it's the classic argument that humans are really good at thinking

and planning and designing.They're not really good at implementing stuff.

we try to be consisting in the implementating of one or two systems, maybe you're lucking. 100 machines,

forget it.

1000 machines, you're probably dead by then anyway.

machines are really good at doing consitant implementation and not very good at desiging or planning.

So separating of duties is a good state of affairs. So that's pretty much the idea.

and as i say, it's like stylesheets of the web, we're trying to separate how we want things to look and

from how we want things to be implemented.

The difference of course between stylesheets and systems is that your web pages aren't usually changing

behind your back.