gnu-arch-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gnu-arch-users] How do do a namespace server right


From: James Blackwell
Subject: Re: [Gnu-arch-users] How do do a namespace server right
Date: Sat, 27 Mar 2004 18:44:48 -0500
User-agent: Mutt/1.5.5.1+cvs20040105i

On Sat, Mar 27, 2004 at 04:17:00PM -0600, Parker, Ron wrote:
> > By making these trackers so simple to implement, they have hit upon a
> > new concept -- decentralized centralization (!). Sure, there's
> > superservers, but there's so many of them, that nobody is in control.
> > 
> > What bittorrent has to teach us is that the problem isn't centralized
> > servers. The problem to avoid is having a centralized server 
> > that is too
> 
> Okay, this is too weird.  After asking you about the bandwidth etc.
> requirements of the supermirror, I got to thinking about bittorrent as it
> might relate to arch, down the road, especially as the supermirror becomes
> overloaded.  It is an interesting idea for a lot of different reasons.
> 
> As I checked into bittorrent, I discovered that it is really designed for
> large files with a fairly concentrated and intense demand period.  For it to
> 
> I haven't had time to get back to investigating it and to think about
> whether or not their might be a way to adapt what bittorrent does to the
> arch environment.

Grin. I'm not proposing that arch archives be pushed via bittorrent. That
just wouldn't work. :) 

What I'm attemping to illustrate is that the problem that we're all
seeking to avoid, the creation of a monopoly for the namespace, is
actually somewhere different than we've been thinking. The problem
is not in having an 'arch namespace server (ANS)'. The problem is that any
such ANS must be a commodity so that anybody can run one out of their
basement.

Does this ASN need to be some sort of specialized server?  HECK NO. We've
learned from both CVS's and subversion's server that specialized servers
are the wrong way to go.  No, the right answer here is to devise a
filesystem protocol that can be served by CFDP[0]. Something like
the way that arch works now. 

For example, what if people could perform the following (I've simplified
the scenario. In a more proper solution, things would be a little more
complex):

1a. "tla lookup --add http://arches.gnuarch.org/serverlist";
   Add http://arches.gnuarch.org/serverlist to
   ~/.arch-params/lookupservers

1b . "tla lookup --add http://evencooler.com/myarches";
  Add http://evencooler.com/myarches to ~/.arch-params/lookupservers

2. "tla register-archive --lookup address@hidden
  Downloads and conflates http://arches.gnuarch.org/serverlist &
  http://evencooler.com/myarches
   
  2a. If only one archive matches, register-archive it.
     
  2b. If more than one archive matches, list all matching archives and 
      tell the user to run register-archive by hand.

3. user creates new archive

4. user registers the archive with a site by visiting a cgi page that
   updates the list of archives in http://arches.gnuarch.org/serverlist
      

[0] CFDP or "Common File Distribution Protocols" - An acronym I just made
up on the spot that refers to servers such as http, ftp, scp, etc.

-- 
James Blackwell          Please do not send me carbon copies of mailing
Smile more!              list posts. Such mail is unsolicited. Thank you!

GnuPG (ID 06357400) AAE4 8C76 58DA 5902 761D  247A 8A55 DA73 0635 7400




reply via email to

[Prev in Thread] Current Thread [Next in Thread]