Essentially all applications will use some sort of computational grid but only some will use high performance
|
"High Performance" can come from several sources
-
Classic "parallel" computing inter-object messages
-
Extreme Quality of Service requirements
-
Large number of (simultaneous or asynchronous) users
-
time-critical/real-time constraints ....
|
Most High Performance grids will come from scaling "ordinary grids" which will use commodity solutions.
-
Building grids is major activity for corporate software systems
|