| 17 | | The existence of many Java based legacy applications implied a need to find an appropriate framework which could provide a similar functionality for parallel Java applications as MPI offers to C/C++ or FORTRAN parallel code. Instead of exploiting existing Java bridges to MPI implementations we decided to use the !ProActive Parallel suite. The library uses the standard Java RMI framework as a portable communication layer. With a reduced set of simple primitives, !ProActive (version 3.9 as used in !QosCosGrid) provides a comprehensive toolkit that simplifies the programming of applications distributed on local area networks, clusters, Internet grids and peer-to-peer intranets for Java-based applications. To satisfy the requirements of complex system simulation applications and users, we developed extensions to the !ProActive library (called QCG-!ProActive) with the following goals: (1) To preserve standard !ProActive library properties (i.e., allow legacy !ProActive applications to be seamlessly ported to !QosCosGrid). (2) To provide end users with a consistent QCG Broker Job Profile schema as a single document used to describe application parameters required for execution as well as resource requirements (in particular network topology and estimated execution time). (3) To prevent end users from the necessity to have direct (i.e., over SSH) access to remote clusters and machines. |
| | 17 | The existence of many Java based legacy applications implied a need to find an appropriate framework which could provide a similar functionality for parallel Java applications as MPI offers to C/C++ or FORTRAN parallel code. Instead of exploiting existing Java bridges to MPI implementations we decided to use the !ProActive Parallel suite. The library uses the standard Java RMI framework as a portable communication layer. With a reduced set of simple primitives, !ProActive (version 3.9 as used in !QosCosGrid) provides a comprehensive toolkit that simplifies the programming of applications distributed on local area networks, clusters, Internet grids and peer-to-peer intranets for Java-based applications. To satisfy the requirements of complex system simulation applications and users, we developed extensions to the !ProActive library (called QCG-!ProActive) with the following goals: |
| | 18 | 1. To preserve standard !ProActive library properties (i.e., allow legacy !ProActive applications to be seamlessly ported to !QosCosGrid). |
| | 19 | 1. To provide end users with a consistent QCG Broker Job Profile schema as a single document used to describe application parameters required for execution as well as resource requirements (in particular network topology and estimated execution time). |
| | 20 | 1. To prevent end users from the necessity to have direct (i.e., over SSH) access to remote clusters and machines. |
