[JDEV] server performance questions

[Corbett J. Klempay]

Ah, sweet.  I just checked it out.  That is a cool idea (the generality
makes it potentially so much more leverageable than a distributed
directory (which is cool in itself)).  Would the hub functionality be
implemented as a separate process, or as a module for a regular jabber
server?  Is the mod_mysql (or mod_whatever_dbserver) module to provide the
logic the hub needs to do its db stuff (in addition to its original
'authenticate users against a db' functionality), or would there have to
be another module for the db being used? (so like if mysql were used,
you'd need both a module to do the user authentication against the db, and
another module to do the hub db stuff)

---
Corbett J. Klempay
Trilogy Software, Inc.
512.532.5176 (W) | 512.750.1372 (C)
corbett.klempay at trilogy.com

On Thu, 20 Jan 2000, temas wrote:

> Enter JUD.  Jabber Universal Data is a way to have a massive distributed
> database of this type of information.  JUD will come to fruition very shortly.
> Until then I suggest reading my introduction to jude at:
> 
> http://docs.jabber.org/proposals/jud/
> 
> That should get you started on the idea of where we want to move on that topic.
> 
> --temas
> 
> On Thu, Jan 20, 2000 at 02:27:16PM -0500, Corbett J. Klempay wrote:
> > Ah wow, that's much better than I was expecting.  How common do you expect
> > Jabber servers to be? (like every ISP running them? lots of
> > personal/college dorm ones?)
> > 
> > Hmm, I'd be curious to know what kind of hardware is running the
> > icq#.mirabilis.com servers.  (I would think ICQ's protocol would be much
> > less server-intensive, though)
> > 
> > Here is a question that is probably been talked about/done...say I'm a
> > Jabber user trying to find a friend...how is directory searching (or the
> > equivalent of) treated? I mean, does the user have to know what server to
> > check, or is there a way to have it mask (from a searching standpoint) the
> > large number of servers, making it appear to the user like they are
> > searching just one large directory?
> > 
> > ---
> > Corbett J. Klempay
> > Trilogy Software, Inc.
> > 512.532.5176 (W) | 512.750.1372 (C)
> > corbett.klempay at trilogy.com
> > 
> > On Thu, 20 Jan 2000, Thomas Charron wrote:
> > 
> > > Quoting "Corbett J. Klempay" <cklempay at chimera.acm.jhu.edu>:
> > > > Something I kept forgetting to ask (but Robert Thompson's last email
> > > > reminded me): how will the jabber server scale to high user loads?  XML
> > > > parsing is slow as hell...it seems like parsing of lots of tags (when you
> > > > have lots of users) will murder a server...
> > > 
> > >   I can't speak for the other developers, but it all depends on what 'high user 
> > > load' is.  That threshold differs from machine to machine.  On a P200 w/64 
> > > Megs, I can connect 1,000 users, and have an average thoughput of 20-40 
> > > messages per second processed overall.  Bear in mind, this is on a machine 
> > > running X, etc, so it's not able to use 100% of the system resources..
> > > 
> > > --- 
> > > Thomas Charron
> > > << Wanted: One decent sig >>
> > > << Preferably litle used  >>
> > > << and stored in garage.  ?>>
> > > 
> > > _______________________________________________
> > > jdev mailing list
> > > jdev at jabber.org
> > > http://mailman.jabber.org/listinfo/jdev
> > > 
> > 
> > 
> > _______________________________________________
> > jdev mailing list
> > jdev at jabber.org
> > http://mailman.jabber.org/listinfo/jdev
> > 
>