Re: [Qemu-devel] FreeOSZoo submission guideline update proposal

[Jean-Michel POURE <jm@poure.com>]

> I'm not sure I'll be able to follow those guides. Most of the OSes I'll be
> installing are for teaching purposes, so graphical logins and so on could
> not be the best choice. That's why I'm even trying to keep the image size
> low.

This is free software, do as you wish, these are only guidelines.

You are free to follow them or not. In understand there are two distinct 
issues: disc space, graphical login vs. text login:

- Disc space needs more discussion. Ronald outlines in a recent email that 
FAT16/32 may be limited to 4 Gb. It means that we will need to publish again 
the 4.3 Gb images like the SuSE. Also, we propably need COW images.

But when installing Fedora Core 2, I had the feeling that COW was slower than 
RAW. Maybe I am wrong and it was a problem with timing. I have to test again.

Does anyone know what is the faster image system, COW or RAW ?

- About graphical login, we can propose distinct download sections.

What do you think?

> The main task of Renzo's idea was preparing something that could be used
> into a laboratory, and that's what I started to do. Of course, the project
> has expanded and so larger images are ok, but a lab could "explode" when
> things start to be so big...

My personal opinion would be to clear the first 10 items of the to-do-list 
quickly, then work on a modern GUI and deliver a large number of guest 
systems.

The to-do-list can be found on the homepage:
http://www.freeoszoo.org

From 1 July to 8 July, the FTP figures were:

                                                  ---- Percent  Of ----
     Archive Section      Files Sent Bytes Sent    Files Sent Bytes Sent
------------------------- ---------- ------------- ---------- ----------
/home/ftp/images                1904   86343862323    74.20      89.97
/home/ftp                         80    9133409529     3.12       9.52
/home/ftp/qemu                   582     494336622    22.68       0.52

On a monthy basis, we will probably reach 10.000 guest systems delivered. In 
the short run, we may need peer-to-peer technologies, like Bittorrent, which 
is already used by RedHat and Mandrake to ship their distros.

My dream would be to deliver one million guest systems on a yearly basis. This 
can only be done if a multi-platform GUI is available. If not, the figures 
will probably be closer to 100.000 guest systems.

Kind regards,
Jean-Michel