Job Management

One of the key uses, if not ``the'' key use, for this nascent ``Grid'' is ``job management.'' By job management we simply mean the stopping, starting, checkpointing, migrating, ...of computer processes on apropos ``Grid'' resources. Imagine writing an accurate simulation of the human brain which models all structure from neurotransmitter to encephalon from rhombencephalon to prosencephalon in BASIC on your Commodore VIC-20 then trying to run this code art only to realize that your VIC-20 is a hobbled gimp with only 3.5K of RAM available to BASIC programs. Naturally, you'd want to execute your beatific code somewhere else, somewhere more appropriate, somewhere where the ``little me'' you've cultured can be set free from the confines of 3.5K. The question is: ``How to release your Galatea from the base stone of 3.5K of RAM?'' The answer is: ``GAT.''

GAT's job management system allows users and application programmers alike to effortlessly launch processes on apropos ``Grid'' resources and subsequently manage such processes. All the application programmer or end user need do is describe in a very general way the software they wish to run and the hardware on which the wish to run on, and GAT does the rest. So, for example, to run this encephalon on an apropos resource all you need do is describe the resource you want to run on, ``A Linux box with 32000000 processors,'' and describe your Galatea, then tell GAT about both. GAT will find such a resource for you, if it exists and you have permissions to use it, then start your artful code on this beast.

• Creating and Destroying Jobs
• Examining Jobs
• Scheduling and Un-Scheduling Jobs
• Stopping Jobs
• Checkpointing Jobs
• Cloning Jobs
• Migrating Jobs