Efficient Aspect Assignment in Heterogeneous Distributed Systems

In recent years, with increasing use of distributed systems, distributed Aspect Oriented Programming (AOP) arouses more interest. The way of distributing aspects over the network can affect the performance of the program. While assigning aspects to hosts, to achieve higher performance, properties of the distributed system and relation between aspects and objects must be taken into consideration. In this context as a solution for the aspect assignment problem in distributed systems we propose two algorithms, namely a Genetic Algorithm (GA) and an A* algorithm. We evaluate the efficiency of these two algorithms by using a simulator on heterogeneous distributed systems, where hosts and communication lines have different properties. It is shown that the GA is more favorable than A* algorithm for larger systems with many nodes while for smaller systems A* may be preferable. The simulation results indicate that the properly assignment of aspects to hosts improves the performance of a distributed aspect oriented program.