The ReMIX cluster is programmed through a framework. The
goal is to simplify the development of indexing algorithms. It
has been designed with 2 major requirements:
simplicity
reutilisability
simplicity comes from a simple master/slave execution model.
Furthermore, the data distribution is explicit and static. Execution, communication
and synchronization are fully handled by the programming environment.
reutilisability both takes place at a software and hardware level.
The programming environment is not linked to any specific application and
doesn't depend of a dedicated platform. Programs can be executed either on a
single PC, a parallel machine (cluster) or a dedicated platform (ReMIX).
This is particularly usefull during the debugging phase.