[AG-TECH] Access Grid Documentation Project

Stuart Levy slevy at ncsa.uiuc.edu
Thu Feb 8 10:47:28 CST 2001


Gregor von Laszewski wrote:

> Hi:
> 
> Note: I am an AG user not an AG developer or document writer.
> 
> I took on monday a very useful tutorial. Nevertheless, in order for me 
> to use the access grid I need an FAQ and scenario kind of quickstart guide.

 [along with lots of other good questions]

> e) How do I use a white board within the AG?

The MICE whiteboard tool "wbd" could be a really handy thing to use.
Besides drawing & typing, it can also efficiently import & share text files
and postscript pictures -- good for those of us who don't spend their
time in powerpoint...  I think this could be a nice addition to the
AG tool suite.  Of course, to be useful, it has to be something that
most everyone has installed, i.e. it'd need to become part of the
standard AG distribution.

  [...]

> also it would be great to have UNIX shell scripts that allow to schedule 
> reservation or free them. This way we could integrate the AG in Problem 
> Solving Environments.
> 
> .e.g.  scheduleAGevent -name "EasyToRememberKeyInsteadOfConfusingRoomName" \
>                                         -start <time> -end <time> 
> -participants ANL,Indiana,Sandia,UC-Flash
> 
> This way I could for example create an event called:
> 
>    "Lysosyme Experiment"
> 
> which has actually a semantic implications to my participants. For all 
> these Mudders out there the semantic meaning is a name of a room, but 
> meeting in a room called sailing is confusing and much more difficult to 
> remember than a term determined from the audience scheduling the event. 
> The software must also provide a lookup function that allows to extract 
> information about the session from a directory service, indicating what 
> capabilities I need to joins the session, ...

This sounds like what sdr, the mbone session directory program, already
does.  It has a nice calendar display to show what's already allocated,
and also gives a visual list of currently-known sessions.  Choose a
session and you see what facilities it requires.  You can find the
chosen multicast addresses (corresponding to AG rooms), but the UI
makes the participant-meaningful name much more prominent.

Differences (that would need to be addressed to be able to use it in the
style that Gregor suggests) seem to be:

  - instead of cooking up random multicast addresses as sdr does now,
	it'd want to select among the handful of rooms.

  - sdr probably doesn't expect to find conflicts -- there's plenty of mcast
	addresses -- but given the limited number of AG rooms,
	it would need to be aware of the possibility.

	(By the way, why the limited number of rooms?  Is it just so that
	everyone can have the presets pre-distributed?  Or to limit the
	aggregate bandwidth?)

  - sdr is intended to be run from its GUI rather than being command-line
	driven.  However, it's Tcl-based, so it might not be
	too much work to provide command-line access to the same
	functions the GUI uses.



More information about the ag-tech mailing list