incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Kevin Kluge <>
Subject RE: [PROPOSAL][RFC] CloudStack for the Apache Incubator
Date Thu, 05 Apr 2012 00:07:51 GMT
Leo, thanks for the feedback, I've put a few replies in line.

-----Original Message-----
From: Leo Simons [] 
Sent: Wednesday, April 04, 2012 9:31 AM
Subject: Re: [PROPOSAL][RFC] CloudStack for the Apache Incubator

On Tue, Apr 3, 2012 at 7:17 PM, Kevin Kluge <> wrote:
> The draft proposal document is available at

Looks great! Also looks like you've got plenty of interest, so that's also great :-)

Some random thoughts:

* It reads like you have some interesting amount of work ahead of you dealing with the legal
side of things, but y'all did a very good of explaining it in the proposal I think.

K> Thanks.  We'll have the source and binary dependencies documented shortly.   There are
a couple of things around BSD-derivative licenses that aren't clear to me.  It's good to have
mentors.  :)

* I think the way that you deal with having debian as a dependency is fine; I am also assuming
there's nothing _that_ fundamental about the use of debian that someone could not replace
it with something different.

K> I'm glad to hear that.   This was a concern point for me.  A previous version of CloudStack
used Fedora for the system VM OS, so it can certainly be changed between Linux flavors.  It
should be possible to switch to a BSD variant, but it would be a fair bit more work, and I'm
not sure what kind of i/o throughput would be achieved on the various hypervisors in that

* I suspect the most challenging GPL dependency could be mysql if you use a lot of mysql-specific
features. Fortunately, Apache APR has already come up with a good model for how to limit the
licensing dependency on GPL databases, while still allowing the 99% of the people that want
to use them to do so (see ).

K> thanks, also very helpful.  CloudStack uses very few, if any, MySQL-specific features.

* I don't think it is *needed* to have a full plan for how you deal with every single dependency
in the proposal. That sounds like a lot of work. Having such a plan is of course good, but
aside from that, you already have a good "plan for a plan" and I think that'd be good enough
to start incubation with.

* Something similar is probably true for the mentioned website(s) -- assuming they don't involve
loads and loads of traffic or scary spam-ridden user forums, it is fine to list them as things-to-do
and then tackle the finer details of it during the incubation process, it sounds like some
of this stuff will take a while. OTOH if you're shipping thousands of VM images out to people
every day, infra may want to say something about that...but that doesn't seem to be the case

K> no, not at all.  The only meaningful traffic for CloudStack occurs from binary download
of the software and the download of the built system VM, which occurs roughly once per installation.



To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message