| 1 |
Globus/Legion and others have identified difficulties stemming from existing batch schedulers such as LSF and Codine which do not immediately support synchronicity between jobs on different machines as is required in metacomputing
-
This needs to be addressed in any deployed system and we would suggest building such schedulers to respect hierarchy with coarse grain scheduling at commodity layer and finer grain decisions in the backend
|
| 2 |
Do you need services such as WebFlow in following foils which supports composition of compound jobs or metaproblems?
|
| 3 |
The Gateway would be distributed with systems deployed at each site
-
The Sandia GrandeGateway would link to local and remote machines
|
