Meeting Notes for 5 November 2008 Technology/Grid Learner Outcome Comments from Marge

Meeting Notes for 5 November 2008

Present: Mihael, Eric, Liz, Tom J., Bob, Tom L.
Absent: Mike W., Randy, Kris, Joao, Dan K.

Technology/Grid Learner Outcome

Most of the telecon involved a discussion of the technology learning outcome. Tom and Liz had sent out a document (attached at bottom of the page) with a suggestion that the technology learning outcome did not represent what happened in the e-Lab and needed revision. It is important to get this LO right because it is central to the question the ELabs project is supposed to answer about whether the use of the grid enhances physics education.

The current LO is:

Demonstrate technical skill in accessing data from the Grid and Grid computing techniques.

They pointed out that in the case of the Cosmic e-Lab, students do not access data on the Grid. The term Grid computing techniques is rather vague.

They proposed having two LO. They had two version for the first LO:

1a) Describe the characteristics of an analysis that would benefit from using grid resources.

Or

1b) Describe and run an analysis job that benefits from using grid resources.

AND

2) Demonstrate an appreciation of the steps necessary in constructing, executing, tracking the products of analysis jobs.

The 2nd LO should require the students to submit jobs to the grid.

Eric responded to their proposal with comments of his own that are included in the document attached.

After some discussion, we concluded that one LO could be:

What is the grid and why it it used?

Mihael pointed out that a valuable comparison between using your laptop machine and using the grid is: using a DC battery to power a device in your home and plugging into the outlet in the wall so that you can have access to much more power. DC – local machine AC – Grid

The grid Café from CERN (link?) has background information on the web that already uses the electrical analogy and we do point to it.

We also talked about how with solar and thermal power, people now can actually give power back to the grid and this goes along with the analogy of people sharing resources for the computational grid.

We thought the students should know about how it is used so that would involve running jobs on the grid.

We discussed the possibility of having an LO that was optional and focused on computation without a tight connection to the actual analysis tools that were being used in the e-Lab. This brought the issue of whether you can have optional LOs and whether it would be difficult for Jean to do evaluation if some students did and other did not do the computational study associated with the LO. The computational activity could involve completely different data and tools from those used by all e-Labs. Students could keep track of the various analyses they submit and the times involved and the resources used and try to plot these results. Eric actually wrote more extensively about this in the attached document. This activity could be part of the Basics; it could be a separate study in the e-Lab, or merely part of a milestone that came before doing the analysis with the data associated with the e-Lab.

Mihael made the point that the choice of infrastructure and computational entities such as transformations and derivations that his group use to implement running jobs on the grid is not necessarily tied to the grid. We should not be expecting students to learn about them. They are not critical to what we are trying to do with the e-Lab. Tom and I had thought they (Virtual Data Techniques) were important features of our use of the grid based mainly on Mike Wilde's emphasis on their power and the fact that we had not been able to actually run jobs on the grid and we had to show some progress.

Both Tom's stressed that whatever students do, it should not be artificial.

Our current tools can use the grid in ways that really require the computational power of the grid. Bob Peterson chooses large data sets and starts jobs that take a long time to run. We could also consider adding parameter sweeps on both the datasets used and the input parameters that define how the analysis is done.

Mihael also emphasized that the grid is good for jobs that can be done in parallel.

Action Item: We concluded that we had to talk to Jean and understand whether we can have optional LOs.

Action Item: Liz will talk to Marge as well.

Comments from Marge

Here is what Marge said after the meeting:

Marge believes that we should have only one LO, not two. She also mentioned that she thought that the LO might actually fit better into the content section rather than the general section so that students could see this in the context of the particular physics they are doing.

We should approach this as if they are scientists and think about what a scientist would want to know when presented with a new tool and whether he would want to use it. For example, if the IT person suggested to a physicist that he should use the grid, he would not have him do computational studies on some unrelated topic, but would probably try running something that physicists already run to show them how it would work on their computers. This is often what was done when physicists were deciding on what mainframe to buy; they will run benchmarks which consist of typical jobs they run.

She suggested that students should know when it is appropriate to use.

(EKQ) -This would suggest one milestone:

Explain what the grid is, why it it used and when it is appropriate to use for their physics analysis.

Maybe we should have a milestone before they do their analysis that introduces them to grid.

She does not think it should be in the Basics because she does not really think it is basic.

-- Main.LizQuigg - 06 Nov 2008
-- Main.EricMyers - 07 Nov 2008

* technology_lo_revised.doc: Proposal for Changes to the Technology LO for I2U2