next up previous contents
Next: Conclusions Up: Analysis and Evaluation of Previous: Results of Measurements

Summary

The proposed methodology and the implementation of the prototype meet most of the requirements. Because of using reflection the concept relies on programming languages which support this technology and the modification of the framework is necessary if another programming language than Java is used.

Complete transparency to the application is however not been achieved. Adaptors hide most of the adaptation mechanism, but are still visible to the application.

The gain of bandwidth depends heavily on the size of the implementation classes and the number of environments. If only small implementation classes for few environments are used, the adaptation version is less efficient than a customized version. The efficency of adaptation increases with the size of implementation classes and the number of environments.


 
Table: Comparison between customized version and dynamic adaptation  
scenario customized version dynamic adaptation
long running methods + ++
big implementation classes - +
few different environments + -

The costs for gained bandwidth is a runtime overhead. This runtime overhead becomes negligible, if an application has a long running time on a host.

Table [*] gives an overview in which cases adaptation may be a better choice than the conventional customized version.


next up previous contents
Next: Conclusions Up: Analysis and Evaluation of Previous: Results of Measurements
Copyright Munich Network Management Team