Re: More information on Cloud spins
----- Original Message -----
From: "Sandro \"red\" Mathys"
<red(a)fedoraproject.org>
To: "Fedora Cloud SIG" <cloud(a)lists.fedoraproject.org>
Sent: Thursday, May 8, 2014 10:30:36 PM
Subject: Re: More information on Cloud spins
On Thu, May 8, 2014 at 10:04 AM, Amit Saha <asaha(a)redhat.com> wrote:
> I happened to come across a ticket that aims to create cloud images for
> spins. I haven't
> followed any discussion at all on that. Where can I learn more about this?
> Basically,
> I am interested to see if a stripped down version of Fedora Scientific can
> be offered
> as a cloud image as well.
Not sure what ticket you are referring to. I don't think we have any
plans to convert any spin into a cloud image. I doubt it would make
any sense for any of the images. I see how you come to think creating
a scientific spin image would be a good thing - many of the tools are
often used in the cloud by scientific researchers (I did cloud at a
technical university, so I know they do for sure). BUT, we want to
keep cloud images small and highly tailored, and therefore a generic
scientific image would just not work. It would need to be tailored for
one specific usage. So for those scientific use cases (minus big data,
for which we'll have an image), we're just going to recommend the base
image. People can either modify that image by adding more software
themselves before uploading it to their cloud or launch it as-is and
install the necessary software afterwards.
If you think one or two highly tailored images would make sense, bring
up your use case (and be ready to work on it / do all the work) and
we're happy to consider it. Personally, I could imagine a R image but
not sure it's worth the effort - the base image is seriously easily
turned into a R image by end users.
Oh, and the ticket you've seen - maybe that's been about the cloud
images having to follow the spin process. That has nothing to do with
the existing spins.
Thanks for the explanation, Sandro. Yes, indeed that is the ticket which made
me think that the specialized desktop spins were being considered in any way
for the cloud images. I can see how that is a problem.
I would think a cloud image with all the Python scientific tools would be a good
offering. Not sure if anyone else think the same.