[AG-TECH] AccessGrid to H.323 (fwd)

[Gregory Goddard]

Forwarding these messages to the list to complete the thread. -g

---------- Forwarded message ----------
Date: Wed, 09 Oct 2002 16:28:51 -0700
From: Philippe Galvez <galvez at hep.caltech.edu>
To: Gregory Goddard <ggoddard at ufl.edu>
Cc: Lawrence A. Rowe <Rowe at bmrc.berkeley.edu>,
     Brent A Nelson <brent at phys.ufl.edu>, Philippe Galvez <galvez at caltech.edu>,
     David.Collados at cern.ch
Subject: Re: [AG-TECH] AccessGrid to H.323

Gregory,

Sorry for delay but I have a lot of meeting these days and left my mailbox
just growing (I''m still in one of them by the way).

To answer: yes you are quite right. The current UF server handles only
traffic VRVS<=> VRVS traffic. But, could handle VRVS<=> AG too if we enable
it.
I always very strongly advise to have in this case 2 machines on separate
segments.
One handling VRVS traffic only and one doing the gateway for "networking"
reasons. Explicitly, the gateway could generate a lot of traffic and since on
the VRVS only conferences their are daily around 20 working worldwide
meetings. I don't want to increase the risk to have a load segment in the
reflector that we penalize all people connected to it.
So, again my advise is that all sites that want to host some AG Venues and
enable the access via VRVS (could be Institutional Venues) provide a PC
running Linux attached to the network as described very well by Larry. The
cost of the PC is usually less than 1K$. So it is probably worthwhile to have
a new one.

For information: The current gateways are: ANL, Internet2 (Ann Arbor), LBNL
and may be soon for NCSA.

Hope this help in understanding.

Regards,
Philippe


Gregory Goddard wrote:

> David, Philippe,
>
> Am I correct in my assumption below?
>
> -g
>
> On Tue, 8 Oct 2002, Greg Goddard wrote:
>
> >
> > UF already has a local VRVS server.
> >
> > Correct me if I'm wrong, but I'm pretty sure that the local VRVS server
> > handles only VRVS<=>VRVS traffic, and _not_ VRVS<=>Access Grid traffic,
> > which is what Brent is trying to accomplish.
> >
> > The VRVS documentation specifically states that the VRVS<=>AccessGrid
> > gateway is provided by ANL, and I suspect this is where the capacity
> > problem to which Brent refers is occurring.
> >
> > Any clarification is appreciated.
> >
> > -g