[AG-TECH] RE: **ACE-grid** ACE Collaboration services - A discussion document

[John Brooke]

Dear Brian,
          The University of Manchester would be very interested in this.
Please avoid September 10th if you can because we are trying to set up a
European project that will link with future AccessGrid developements and
we are having a consortiu project meeting on that day. I think that the
moves to defining collaboration services are very important to us and we
would like to be involved. Details of our proposal (called REALISTE) are
at

http://www.realiste.org

We assume that you are referring to a meeting via AccessGrid.
                    John Brooke

On Wed, 27 Aug 2003, Brian Corrie wrote:

> Hello all,
>
> As Jason has said (see email below), we are organizing an AG meeting to
> discuss ACE collaboration services, why they are important, and how they
> should be described. This is a continutation of the "ACE collaboration
> services" discussion that we had at GGF in Seattle. As a point of discussion,
> here is a brief position statement from my point of view... This is based on
> the GGF discussion from Seattle (powerpoint provided).
> http://calder.ncsa.uiuc.edu/ACE-grid/2003-06-24_GGF8/GGF8-corrie.ppt
>
> We are planning on having a meeting either September 10, 11, or 12. Please
> let us know if you are interested in participating and if so which
> dates/times are best.
>
> I have cross posted this to the ag-tech mailing list as I think there will be
> people there that are interested in this area that may not be on the ACE
> list...
>
> >From our point of view, providing a succesful collaboration requires a number
> of steps. One has to:
>
> 1) Understand the collaboration task being undertaken
> 2) Understand the needs required to accomplish the task
> 3) Determine the collaboration services required to meet those needs
> 4) Determine how to deliver those services most effectively on the technology
> available to the user.
>
> My goal in moving this discussion forward is to define what we mean by a
> collaboration service so that we can perform the mapping from needs (2 above)
> to services (3 above) to technologies (4 above).
>
> To accomplish the mapping from need to service the services have to be well
> defined. They should be standards based and not application based if
> possible. We need to be able to identify what services meet the needs of the
> collaboration task and be able to combine them in a very flexible manner to
> accomplish that task. That means that services need to work together to
> accomplish an overall goal that can be quite complex. To determine how to do
> that one needs to know alot about the "capabilities" of the services. It is
> probably necessary to classify services into groups that satisfy a given need
> (classes might be audio, video, shared presentation, application sharing,
> floor control, turn taking, visualization, ...). For example, a specific
> collaboration task may NEED audio, video, and shared presentation. There may
> be a number of services that provide a solution for each need. Vic,
> netmeeting, realvideo, and quicktime all satisfy the need to have video. The
> NEED has certain requirements (interactive video) and services have certain
> capabilities (vic is interactive, realvideo is not). If one knows the needs
> of the task and the capabilities of the services we can map from need to
> service to come up with a set of services that satisfy the needs of a
> collaboration (vic satisfies the interactive video need while realvideo does
> not). Thus far we are talking at an abstract level that is focused on
> accomplishing the task at hand as succesfully as possible. We call this
> "Quality of Experience" or QoE.
>
> Once a set of services is identified for a task, we need to map those
> services onto a set of technologies to actually deliver the collaboration. To
> do this, we need to know both the resources required for the services and the
> capabilities of the devices that the services are going to be running on.
> Running vic at 256Kb/s h261 codec requires 256kb/s bandwidth per video stream
> and 50%CPU + 10% per video stream (I know, % CPU is not the best metric in
> the world 8-). Machine A is connected to the other collaborators over the
> research network at 10 Mbps and has a video capture capability. If we know
> this information for each service in a collaboration and each machine being
> used we can make intelligent aggregate decisions at the task level about what
> machines should run which services and more importantly when and how the
> system should compromise on specific needs to make the collaboration as high
> quality as possible (degrade the video quality rather than have poor audio in
> the collaboration).
>
> To perform any of these mappings it is critical to define what a
> collaboration service is and what its capabilities and attributes are. More
> importantly, if we can describe service capabilities in a well defined
> manner, then a true service model can be built in which services can be used
> across multiple systems but decisions can be made about the collaboration
> task at the task level.
>
> So what are the questions we need to ask (from the powerpoint)?
>
> How do we classify collaboration services (or do we need to)?
> How do we describe collaboration services?
> 	- What class of service is it?
> 	- What are its capabilities (what can it do)?
> 	- How does it do it (protocols, standards)?
> 	- What does it require (memory, bandwidth, CPU)?
> How do we register collaboration services (registration and discovery)?
> How do we instantiate services (service creation)?
>
> That is our view of collaboration services and how they should be described.
> I hope that it is relatively clear. I would welcome any discussion on this
> topic and I hope to see some of you at the AG meeting next month...
>
> Cheers,
>
> Brian
>
>
> > -----Original Message-----
> > From: Jason Leigh [mailto:spiff at evl.uic.edu]
> > Sent: Wednesday, August 13, 2003 5:37 PM
> > To: 'Ace'
> > Subject: RE: **ACE-grid** ACE Collaboration services
> >
> >
> > Looks like there is a conflict with the AG2 tutorial. How
> > about Sept 10,
> > 11 or 12?
> >
> > Jason
> >
> >  ----------------
> > Jason Leigh, PhD
> > Associate Professor, Computer Science Department
> > spiff at evl.uic.edu              Electronic Visualization Lab (M/C 152)
> > EVL Phone (312) 996-3002         University of Illinois at Chicago
> > EVL FAX   (312) 413-7585         851 S. Morgan St. Room 1120 SEO
> > http://www.evl.uic.edu/spiff     Chicago, IL 60607-7053
> >
> >
> > > -----Original Message-----
> > > From: owner-ace-grid at gridforum.org
> > > [mailto:owner-ace-grid at gridforum.org] On Behalf Of Jason Leigh
> > > Sent: Wednesday, August 13, 2003 3:19 PM
> > > To: Ace
> > > Subject: **ACE-grid** ACE Collaboration services
> > >
> > >
> > > Hi
> > > Brian and I are planning to hold an AG meeting to elaborate on ACE
> > > collaboration services that was discussed at GGF8. For those
> > > interested,
> > > please indicate your prefered dates for the meeting: Sept 3
> > or 4 or 5.
> > >
> > > Here are notes from the last discussion:
> > > http://calder.ncsa.uiuc.edu/ACE-grid/2003-06-24_GGF8/GGF8-corrie.ppt
> > >
> > > Thanks
> > > Jason
> > >
> > >  ----------------
> > > Jason Leigh, PhD
> > > Associate Professor, Computer Science Department
> > > spiff at evl.uic.edu              Electronic Visualization Lab
> > (M/C 152)
> > > EVL Phone (312) 996-3002         University of Illinois at Chicago
> > > EVL FAX   (312) 413-7585         851 S. Morgan St. Room 1120 SEO
> > > http://www.evl.uic.edu/spiff     Chicago, IL 60607-7053
> > >
> >
> >
>