The remote execution of functions for scientific applications creates interesting possibilities of software architectures. For example: it is possible to conceive software in a tier containing only graphical presentation and interface, and other tier containing numeric execution. An architecture with tier isolation may produce several very desirable consequences, including the improvement of the maintainability of the code, enhancement of the intellectual property protection, improved flexibility to handle the technological substitution of hardware, the independence of type of client for the numeric functions, and others.
In this work, it is proposed a software architecture - named NumerusCloud - that uses cloud computing technologies to implement with 3 tiers a system that allows the automation of deployment of numeric functions. The proposed architecture offers important advantages over the known techniques of remote execution, including the isolation of the tiers of the deployment agent and the tier of the author of numerical task.
The proposed architecture was implemented, and as usage example it was deployed a numeric function to calculate the clustering problem (a function where intellectual restrictions apply), using 2 types of clients in presentation tier - one as a GUI for PC and other for mobile (Android).
.png)
.png)
