[Debtorrent-devel] Fwd: BitTorrent Protocol Expansion (Google SoC)

Cameron Dale camrdale at gmail.com
Fri Apr 13 01:45:35 UTC 2007


---------- Forwarded message ----------
From: Anthony Towns <aj at azure.humbug.org.au>
Date: Mar 26, 2007 10:13 PM
Subject: Re: BitTorrent Protocol Expansion (Google SoC)
To: Cameron Dale <camrdale at gmail.com>


On Mon, Mar 26, 2007 at 05:25:05PM -0700, Cameron Dale wrote:
> http://www.cs.sfu.ca/~camerond/personal/GoogleSoCDebian.html

Some comments:

    This project proposes to create a backend or proxy to the Debian
    package distribution tool apt, which will allow it to download
    packages from other users of Debian in a BitTorrent-like manner,
    thus reducing the costs incurred by the [Debian project].

I'm not sure it's the costs to the Debian project that matter, so much
as the costs for random people to setup Debian archives that get used
by a lot of people.

I would think a standalone mirroring app (like apt-proxy) is likely far
easier than integrating it into apt.

    These costs are currently being reduced by the voluntary mirroring
    system Debian uses to help distribute packages. This system comes
    with some drawbacks though [...]

One you didn't mention was that it's slow to react to changes in the
mirror network -- users have to change their sources list by hand if
they need to switch to a different mirror, and admins have to keep their
upstream (and sometimes downstream) mirrors organised by hand too.

    Finally, the archive is frequently updated, though only a very small
    portion of it at a time.

See http://ftp-master.debian.org/size-quarter.png for a graph of the
amount that changes daily. We seem to be getting about 1GB or 2GB per day,
compared to a current total archive size of 200GB or so I guess.

    running many copies of a modified BitTorrent client on the PlanetLab
    research testbed.

What's the client? BitTorrent itself or BitTornado or..?

> I've tried to make the application reflect the reduced scope of the
> project, in the hopes of increasing the probability of success. In
> particular, I think the requirement for the BitTorrent system to notify
> a user of updated packages leads to a lot of complexity early on (at
> least it seemed to from the emails you sent), and is not really the
> focus of the project. I've therefore moved it to an optional section at
> the end.

Good-o.

> A contact schedule is not a problem, I have no big plans for the summer,
> nor do I foresee losing Internet access. ;)

Excellent. :)

Cheers,
aj


-----BEGIN PGP SIGNATURE-----

iD8DBQFGCKgJOxe8dCpOPqoRApZhAJ4sYC/UbBNjSAcXVxiNgZhx2e3LxACgm7lJ
5f3HrFJFm/5Pi0xeb7d7pLo=
=RZJn
-----END PGP SIGNATURE-----



More information about the Debtorrent-devel mailing list