Changes between Version 2 and Version 3 of WorkShop09/ScientistUsability


Ignore:
Timestamp:
Oct 29, 2009, 3:38:28 PM (14 years ago)
Author:
Nicolas
Comment:

Format as bullet list. Not sure if it's the most readable formatting but...

Legend:

Unmodified
Added
Removed
Modified
  • WorkShop09/ScientistUsability

    v2 v3  
    1 {{{
    2 Toni
    3 - Extensions - requirements are training
    4 - perspective - web interface is not general enough
    5 - GPU Grid's Soap interface is general enough
    6 David
    7 - User Defined in this discussion is the Scientist
    8 - Different categories of users
    9 - Envisage a context where random volunteers can submit
    10 - Need to control the scripts where users can inject viruses
    11 Toni
    12 - Users should not have ability to upload applications - this is a security risk
    13 - Applications that accept scripts should be tightly restricted
    14 Christian
    15 - Extend template files w/ additional text.
    16 - Soap can be used for a web page where scientist selects for
    17 application to submit
    18 - Scientist must provide a template
    19 Toni
    20 - Webpage for submitting:
    21 1) Does not lend itself to batches
    22 2) Output would not be able to retrieve on webpage
    23 Christian
    24 - Suggested a structure where input is packed in an archive, uploaded
    25 to the server, uncompressed and used by job creation script
    26 Zoltan
    27 -Change Execution soap website
    28 -Client executes application
    29 -Master feeds the work
    30 Toni
    31 -Described how GPUGrid accounts for the priority of workunits. High
    32 priority starves the lower priority
    33 -Will release the Perl code
    34 -Does not deal w/ scheduler code
    35 -One should have a scientist's table that tracks various project
    36 scientist information relevant to submissions
    37 David
    38 - Many options [for scientist's table], eg accounting, controlling
    39 latency, throughput
    40 Toni
    41 -Not many scientists on GPU grid currently
    42 Micheal Weber
    43 -Leiden Classical has a drop-down menu
    44 -submissions interface w/ limited characters
    45 -Registers the heavy users
    46 David
    47 -Is Leiden a generic solution?
    48 Christian
    49 -Leiden uses a separate configuration file
    50 Toni
    51 -Computes the has of the file, avoid copying files, conserve space,
    52 use hard links in the download directory
    53 Kevin
    54 -There should be a common mechanism to submit
    55 Zoltan
    56 -How to submit work remotely
    57 David
    58 -It would be good to have a standardized way
    59 Toni
    60 -Use MD5 checksum on the input files, store the hashes on the server
    61 Jack
    62 -What if there were a script to generate the templates, where user is
    63 prompted for various parameters?
    64 David
    65 -Not the bottleneck for scientists
    66 -Need to move the WU Generator and Assimilator off the server
    67 -Keep the things we trust stay on the server
    68 
    69 -- Jack http://drugdiscoveryathome.com http://hydrogenathome.org
    70 
    71 }}}
     1 - Toni
     2   - Extensions - requirements are training
     3   - perspective - web interface is not general enough
     4   - GPU Grid's Soap interface is general enough
     5 - David
     6   - User Defined in this discussion is the Scientist
     7   - Different categories of users
     8   - Envisage a context where random volunteers can submit
     9   - Need to control the scripts where users can inject viruses
     10 - Toni
     11   - Users should not have ability to upload applications - this is a security risk
     12   - Applications that accept scripts should be tightly restricted
     13 - Christian
     14   - Extend template files w/ additional text.
     15   - Soap can be used for a web page where scientist selects for application to submit
     16   - Scientist must provide a template
     17 - Toni
     18   - Webpage for submitting:
     19     1. Does not lend itself to batches
     20     2. Output would not be able to retrieve on webpage
     21 - Christian
     22   - Suggested a structure where input is packed in an archive, uploaded to the server, uncompressed and used by job creation script
     23 - Zoltan
     24   - Change Execution soap website
     25   - Client executes application
     26   - Master feeds the work
     27 - Toni
     28   - Described how GPUGrid accounts for the priority of workunits. High priority starves the lower priority
     29   - Will release the Perl code
     30   - Does not deal w/ scheduler code
     31   - One should have a scientist's table that tracks various project scientist information relevant to submissions
     32 - David
     33   - Many options [for scientist's table], eg accounting, controlling latency, throughput
     34 - Toni
     35   - Not many scientists on GPU grid currently
     36 - Micheal Weber
     37   - Leiden Classical has a drop-down menu
     38   - submissions interface w/ limited characters
     39   - Registers the heavy users
     40 - David
     41   - Is Leiden a generic solution?
     42 - Christian
     43   - Leiden uses a separate configuration file
     44 - Toni
     45   - Computes the has of the file, avoid copying files, conserve space,
     46   use hard links in the download directory
     47 - Kevin
     48   - There should be a common mechanism to submit
     49 - Zoltan
     50   - How to submit work remotely
     51 - David
     52   - It would be good to have a standardized way
     53 - Toni
     54   - Use MD5 checksum on the input files, store the hashes on the server
     55 - Jack
     56   - What if there were a script to generate the templates, where user is
     57   prompted for various parameters?
     58 - David
     59   - Not the bottleneck for scientists
     60   - Need to move the WU Generator and Assimilator off the server
     61   - Keep the things we trust stay on the server